lang: Add internal, resource specific edges

This adds the ability to specify internal, resource specific edges, with
and without notifications. We use the special words: "Notify", "Before",
"Listen", and "Depend". They must have the first character capitalized.
They also support the "elvis" operator.

.editorconfig

@@ -12,8 +12,14 @@ end_of_line = lf
 insert_final_newline = true
 trim_trailing_whitespace = true
 
+[*.sh]
+indent_style = tab
+
 [*.go]
 indent_style = tab
 
 [Makefile]
 indent_style = tab
+
+[*.mcl]
+indent_style = tab

.github/PULL_REQUEST_TEMPLATE.md

@@ -1,20 +1,30 @@
 ## Tips:
 
+* please read the style guide before submitting your patch:
+[docs/style-guide.md](../docs/style-guide.md)
+
 * commit message titles must be in the form:
+
 ```topic: Capitalized message with no trailing period```
+
 or:
+
 ```topic, topic2: Capitalized message with no trailing period```
 
 * golang code must be formatted according to the standard, please run:
+
 ```
 make gofmt		# formats the entire project correctly
 ```
+
 or format a single golang file correctly:
+
 ```
 gofmt -w yourcode.go
 ```
 
 * please rebase your patch against current git master:
+
 ```
 git checkout master
 git pull origin master
@@ -25,6 +35,7 @@ hub pull-request	# or submit with the github web ui
 ```
 
 * after a patch review, please ping @purpleidea so we know to re-review:
+
 ```
 # make changes based on reviews...
 git add -p		# add new changes

.github/settings.yml

@@ -0,0 +1,96 @@
+# These settings are synced to GitHub by https://probot.github.io/apps/settings/
+
+repository:
+  # See https://developer.github.com/v3/repos/#edit for all available settings.
+
+  # The name of the repository. Changing this will rename the repository
+  name: mgmt
+
+  # A short description of the repository that will show up on GitHub
+  description: Next generation distributed, event-driven, parallel config management!
+
+  # A URL with more information about the repository
+  homepage: https://purpleidea.com/tags/mgmtconfig/
+
+  # A comma-separated list of topics to set on the repository
+  topics: golang, go, configuration-management, config-management, devops, etcd, distributed-systems, graph-theory, choreography
+
+  # Either `true` to make the repository private, or `false` to make it public.
+  private: false
+
+  # Either `true` to enable issues for this repository, `false` to disable them.
+  has_issues: true
+
+  # Either `true` to enable projects for this repository, or `false` to disable them.
+  # If projects are disabled for the organization, passing `true` will cause an API error.
+  has_projects: false
+
+  # Either `true` to enable the wiki for this repository, `false` to disable it.
+  has_wiki: false
+
+  # Either `true` to enable downloads for this repository, `false` to disable them.
+  has_downloads: true
+
+  # Updates the default branch for this repository.
+  default_branch: master
+
+  # Either `true` to allow squash-merging pull requests, or `false` to prevent
+  # squash-merging.
+  allow_squash_merge: false
+
+  # Either `true` to allow merging pull requests with a merge commit, or `false`
+  # to prevent merging pull requests with merge commits.
+  allow_merge_commit: false
+
+  # Either `true` to allow rebase-merging pull requests, or `false` to prevent
+  # rebase-merging.
+  allow_rebase_merge: true
+
+# Labels: define labels for Issues and Pull Requests (in alphabetical order)
+labels:
+  - name: bug
+    color: fc2929
+  - name: confirmed
+    color: d93f0b
+  - name: design
+    color: 5319e7
+  - name: duplicate
+    color: cccccc
+  - name: enhancement
+    color: 84b6eb
+  - name: good first issue
+    color: 7057ff
+  - name: help wanted
+    color: 159818
+  - name: invalid
+    color: e6e6e6
+  - name: mgmtlove
+    color: e11d21
+  - name: question
+    color: cc317c
+  - name: wontfix
+    color: ffffff
+#  - name: first-timers-only
+#    # include the old name to rename an existing label
+#    oldname: Help Wanted
+
+# Collaborators: give specific users access to this repository.
+#collaborators:
+#  - username: purpleidea
+#    # Note: Only valid on organization-owned repositories.
+#    # The permission to grant the collaborator. Can be one of:
+#    # * `pull` - can pull, but not push to or administer this repository.
+#    # * `push` - can pull and push, but not administer this repository.
+#    # * `admin` - can pull, push and administer this repository.
+#    permission: push
+
+#  - username: hubot
+#    permission: pull
+
+# NOTE: The APIs needed for teams are not supported yet by GitHub Apps
+# https://developer.github.com/v3/apps/available-endpoints/
+#teams:
+#  - name: core
+#    permission: admin
+#  - name: docs
+#    permission: push

.gitignore

@@ -6,7 +6,11 @@
 old/
 tmp/
 *_stringer.go
+bindata/*.go
 mgmt
 mgmt.static
+# crossbuild artifacts
+build/mgmt-*
 mgmt.iml
 rpmbuild/
+*.deb

.travis.yml

@@ -1,25 +1,40 @@
 language: go
+os:
+  - linux
 go:
-  - 1.8.x
   - 1.9.x
+  - 1.10.x
   - tip
 go_import_path: github.com/purpleidea/mgmt
 sudo: true
 dist: trusty
 before_install:
-  - sudo apt update
+  # as per a number of comments online, this might mitigate some flaky fails...
+  - if [[ "$TRAVIS_OS_NAME" != "osx" ]]; then sudo apt-key adv --keyserver hkp://keyserver.ubuntu.com:80 --recv 0C49F3730359A14518585931BC711F9BA15703C6; fi
+  # apt update tends to be flaky in travis, retry up to 3 times on failure
+  # https://docs.travis-ci.com/user/common-build-problems/#travis_retry
+  - if [[ "$TRAVIS_OS_NAME" != "osx" ]]; then travis_retry travis_retry sudo apt update; fi
+  - git config remote.origin.fetch "+refs/heads/*:refs/remotes/origin/*"
   - git fetch --unshallow
 install: 'make deps'
 script: 'make test'
 matrix:
-  fast_finish: true
+  fast_finish: false
   allow_failures:
+    - go: 1.10.x
     - go: tip
-    - go: 1.9.x
+    - os: osx
+  # include only one build for osx for a quicker build as the nr. of these runners are sparse
+  include:
+    - os: osx
+      go: 1.9.x
+
+# the "secure" channel value is the result of running: ./misc/travis-encrypt.sh
+# with a value of: irc.freenode.net#mgmtconfig to eliminate noise from forks...
 notifications:
   irc:
     channels:
-      - "irc.freenode.net#mgmtconfig"
+      - secure: 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
     template:
       - "%{repository} (%{commit}: %{author}): %{message}"
       - "More info : %{build_url}"
@@ -29,6 +44,6 @@ notifications:
     skip_join: false
   email:
     recipients:
-      - travis-ci@shubin.ca
+      - secure: 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
     on_failure: change
     on_success: change

AUTHORS

@@ -1,10 +1,12 @@
 This is a list of authors/contributors to the mgmt project.
-If you're a contributor, please send a patch with your name.
+If you're a core contributor, we might ask you to send a patch with your name.
 If you appreciate the work of one of the contributors, thank them a beverage!
 For a more exhaustive list please run: git log --format='%aN' | sort -u
 This list is sorted alphabetically by first name.
 
 Felix Frank
 James Shubin
+Johan Bloemberg
+Jonathan Gold
 Julien Pivotto
 Paul Morgan

COPYRIGHT

@@ -1,5 +1,5 @@
 Mgmt
-Copyright (C) 2013-2017+ James Shubin and the project contributors
+Copyright (C) 2013-2018+ James Shubin and the project contributors
 Written by James Shubin <james@shubin.ca> and the project contributors
 
 This program is free software: you can redistribute it and/or modify

Makefile

@@ -1,5 +1,5 @@
 # Mgmt
-# Copyright (C) 2013-2017+ James Shubin and the project contributors
+# Copyright (C) 2013-2018+ James Shubin and the project contributors
 # Written by James Shubin <james@shubin.ca> and the project contributors
 #
 # This program is free software: you can redistribute it and/or modify
@@ -16,13 +16,15 @@
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 
 SHELL = /usr/bin/env bash
-.PHONY: all art cleanart version program path deps run race generate build clean test gofmt yamlfmt format docs rpmbuild mkdirs rpm srpm spec tar upload upload-sources upload-srpms upload-rpms copr
-.SILENT: clean
+.PHONY: all art cleanart version program lang path deps run race bindata generate build build-debug crossbuild clean test gofmt yamlfmt format docs rpmbuild mkdirs rpm srpm spec tar upload upload-sources upload-srpms upload-rpms copr
+.SILENT: clean bindata
+
+# a large amount of output from this `find`, can cause `make` to be much slower!
+GO_FILES := $(shell find * -name '*.go' -not -path 'old/*' -not -path 'tmp/*')
 
 SVERSION := $(or $(SVERSION),$(shell git describe --match '[0-9]*\.[0-9]*\.[0-9]*' --tags --dirty --always))
 VERSION := $(or $(VERSION),$(shell git describe --match '[0-9]*\.[0-9]*\.[0-9]*' --tags --abbrev=0))
 PROGRAM := $(shell echo $(notdir $(CURDIR)) | cut -f1 -d"-")
-OLDGOLANG := $(shell go version | grep -E 'go1.3|go1.4')
 ifeq ($(VERSION),$(SVERSION))
 	RELEASE = 1
 else
@@ -38,8 +40,12 @@ USERNAME := $(shell cat ~/.config/copr 2>/dev/null | grep username | awk -F '='
 SERVER = 'dl.fedoraproject.org'
 REMOTE_PATH = 'pub/alt/$(USERNAME)/$(PROGRAM)'
 ifneq ($(GOTAGS),)
-    BUILD_FLAGS = -tags '$(GOTAGS)'
+	BUILD_FLAGS = -tags '$(GOTAGS)'
 endif
+GOOSARCHES ?= linux/amd64 linux/ppc64 linux/ppc64le linux/arm64 darwin/amd64
+
+GOHOSTOS = $(shell go env GOHOSTOS)
+GOHOSTARCH = $(shell go env GOHOSTARCH)
 
 default: build
 
@@ -101,40 +107,74 @@ run:
 race:
 	find . -maxdepth 1 -type f -name '*.go' -not -name '*_test.go' | xargs go run -race -ldflags "-X main.program=$(PROGRAM) -X main.version=$(SVERSION)"
 
+# generate go files from non-go source
+bindata:
+	@echo "Generating: bindata..."
+	$(MAKE) --quiet -C bindata
+
 generate:
 	go generate
 
-build: $(PROGRAM)
+lang:
+	@# recursively run make in child dir named lang
+	@echo "Generating: lang..."
+	$(MAKE) --quiet -C lang
 
-$(PROGRAM): main.go
-	@echo "Building: $(PROGRAM), version: $(SVERSION)..."
-ifneq ($(OLDGOLANG),)
-	@# avoid equals sign in old golang versions eg in: -X foo=bar
-	time go build -ldflags "-X main.program $(PROGRAM) -X main.version $(SVERSION)" -o $(PROGRAM) $(BUILD_FLAGS);
-else
-	time go build -i -ldflags "-X main.program=$(PROGRAM) -X main.version=$(SVERSION)" -o $(PROGRAM) $(BUILD_FLAGS);
-endif
+# build a `mgmt` binary for current host os/arch
+$(PROGRAM): build/mgmt-${GOHOSTOS}-${GOHOSTARCH}
+	cp $< $@
 
-$(PROGRAM).static: main.go
+$(PROGRAM).static: $(GO_FILES)
 	@echo "Building: $(PROGRAM).static, version: $(SVERSION)..."
 	go generate
-ifneq ($(OLDGOLANG),)
-	@# avoid equals sign in old golang versions eg in: -X foo=bar
-	go build -a -installsuffix cgo -tags netgo -ldflags '-extldflags "-static" -X main.program $(PROGRAM) -X main.version $(SVERSION)' -o $(PROGRAM).static $(BUILD_FLAGS);
-else
-	go build -a -installsuffix cgo -tags netgo -ldflags '-extldflags "-static" -X main.program=$(PROGRAM) -X main.version=$(SVERSION)' -o $(PROGRAM).static $(BUILD_FLAGS);
-endif
+	go build -a -installsuffix cgo -tags netgo -ldflags '-extldflags "-static" -X main.program=$(PROGRAM) -X main.version=$(SVERSION) -s -w' -o $(PROGRAM).static $(BUILD_FLAGS);
+
+build: LDFLAGS=-s -w
+build: $(PROGRAM)
+
+build-debug: LDFLAGS=
+build-debug: $(PROGRAM)
+
+# pattern rule target for (cross)building, mgmt-OS-ARCH will be expanded to the correct build
+# extract os and arch from target pattern
+GOOS=$(firstword $(subst -, ,$*))
+GOARCH=$(lastword $(subst -, ,$*))
+build/mgmt-%: $(GO_FILES) | bindata lang
+	@echo "Building: $(PROGRAM), os/arch: $*, version: $(SVERSION)..."
+	@# reassigning GOOS and GOARCH to make build command copy/pastable
+	time env GOOS=${GOOS} GOARCH=${GOARCH} go build -i -ldflags "-X main.program=$(PROGRAM) -X main.version=$(SVERSION) ${LDFLAGS}" -o $@ $(BUILD_FLAGS);
+
+# create a list of binary file names to use as make targets
+crossbuild_targets = $(addprefix build/mgmt-,$(subst /,-,${GOOSARCHES}))
+crossbuild: ${crossbuild_targets}
 
 clean:
+	$(MAKE) --quiet -C bindata clean
+	$(MAKE) --quiet -C lang clean
 	[ ! -e $(PROGRAM) ] || rm $(PROGRAM)
 	rm -f *_stringer.go	# generated by `go generate`
 	rm -f *_mock.go		# generated by `go generate`
+	# crossbuild artifacts
+	rm -f build/mgmt-*
 
-test:
+test: build
 	./test.sh
 
+# create all test targets for make tab completion (eg: make test-gofmt)
+test_suites=$(shell find test/ -maxdepth 1 -name test-* -exec basename {} .sh \;)
+# allow to run only one test suite at a time
+${test_suites}: test-%: build
+	./test.sh $*
+
+# targets to run individual shell tests (eg: make test-shell-load0)
+test_shell=$(shell find test/shell/ -maxdepth 1 -name "*.sh" -exec basename {} .sh \;)
+$(addprefix test-shell-,${test_shell}): test-shell-%: build
+	./test/test-shell.sh "$*.sh"
+
 gofmt:
-	find . -maxdepth 3 -type f -name '*.go' -not -path './old/*' -not -path './tmp/*' -exec gofmt -w {} \;
+	# TODO: remove gofmt once goimports has a -s option
+	find . -maxdepth 6 -type f -name '*.go' -not -path './old/*' -not -path './tmp/*' -not -path './vendor/*' -exec gofmt -s -w {} \;
+	find . -maxdepth 6 -type f -name '*.go' -not -path './old/*' -not -path './tmp/*' -not -path './vendor/*' -exec goimports -w {} \;
 
 yamlfmt:
 	find . -maxdepth 3 -type f -name '*.yaml' -not -path './old/*' -not -path './tmp/*' -not -path './omv.yaml' -exec ruby -e "require 'yaml'; x=YAML.load_file('{}').to_yaml.each_line.map(&:rstrip).join(10.chr)+10.chr; File.open('{}', 'w').write x" \;
@@ -277,4 +317,27 @@ upload-rpms: rpmbuild/RPMS/ rpmbuild/RPMS/SHA256SUMS rpmbuild/RPMS/SHA256SUMS.as
 copr: upload-srpms
 	./misc/copr-build.py https://$(SERVER)/$(REMOTE_PATH)/SRPMS/$(SRPM_BASE)
 
+#
+#	deb build
+#
+
+deb:
+	./misc/gen-deb-changelog-from-git.sh
+	dpkg-buildpackage
+	# especially when building in Docker container, pull build artifact in project directory.
+	cp ../mgmt_*_amd64.deb ./
+	# cleanup
+	rm -rf debian/mgmt/
+
+build_container:
+	docker build -t purpleidea/mgmt-build -f docker/Dockerfile.build .
+	docker run -td --name mgmt-build purpleidea/mgmt-build
+	docker cp mgmt-build:/root/gopath/src/github.com/purpleidea/mgmt/mgmt .
+	docker build -t purpleidea/mgmt -f docker/Dockerfile.static .
+	docker rm mgmt-build || true
+
+clean_container:
+	docker rmi purpleidea/mgmt-build
+	docker rmi purpleidea/mgmt
+
 # vim: ts=8

README.md

@@ -9,6 +9,7 @@
 [![Jenkins](https://img.shields.io/badge/jenkins-status-brightgreen.svg?style=flat-square)](https://ci.centos.org/job/purpleidea-mgmt/)
 
 ## Community:
+
 Come join us in the `mgmt` community!
 
 | Medium | Link |
@@ -18,70 +19,65 @@ Come join us in the `mgmt` community!
 | Mailing list | [mgmtconfig-list@redhat.com](https://www.redhat.com/mailman/listinfo/mgmtconfig-list) |
 
 ## Status:
-Mgmt is a fairly new project.
-We're working towards being minimally useful for production environments.
-We aren't feature complete for what we'd consider a 1.x release yet.
-With your help you'll be able to influence our design and get us there sooner!
+
+Mgmt is a next generation automation tool. It has similarities to other tools in
+the configuration management space, but has a fast, modern, distributed systems
+approach. The project contains an engine and a language.
+[Please have a look at an introductory video or blog post.](docs/on-the-web.md)
+
+Mgmt is a fairly new project. It is usable today, but not yet feature complete.
+With your help you'll be able to influence our design and get us to 1.0 sooner!
+Interested developers should read the [quick start guide](docs/quick-start-guide.md).
 
 ## Documentation:
+
 Please read, enjoy and help improve our documentation!
 
 | Documentation | Additional Notes |
 |---|---|
-| [general documentation](docs/documentation.md) | for everyone |
 | [quick start guide](docs/quick-start-guide.md) | for mgmt developers |
+| [frequently asked questions](docs/faq.md) | for everyone |
+| [general documentation](docs/documentation.md) | for everyone |
+| [language guide](docs/language-guide.md) | for everyone |
+| [function guide](docs/function-guide.md) | for mgmt developers |
 | [resource guide](docs/resource-guide.md) | for mgmt developers |
+| [style guide](docs/style-guide.md) | for mgmt developers |
 | [godoc API reference](https://godoc.org/github.com/purpleidea/mgmt) | for mgmt developers |
 | [prometheus guide](docs/prometheus.md) | for everyone |
 | [puppet guide](docs/puppet-guide.md) | for puppet sysadmins |
+| [development](docs/development.md) | for mgmt developers |
 
 ## Questions:
+
 Please ask in the [community](#community)!
-If you have a well phrased question that might benefit others, consider asking it by sending a patch to the documentation [FAQ](https://github.com/purpleidea/mgmt/blob/master/docs/documentation.md#usage-and-frequently-asked-questions) section. I'll merge your question, and a patch with the answer!
+If you have a well phrased question that might benefit others, consider asking
+it by sending a patch to the [FAQ](docs/faq.md) section. I'll merge your
+question, and a patch with the answer!
 
 ## Roadmap:
+
+Feel free to grab one of the straightforward [#mgmtlove](https://github.com/purpleidea/mgmt/labels/mgmtlove)
+issues if you're a first time contributor to the project or if you're unsure
+about what to hack on!
 Please see: [TODO.md](TODO.md) for a list of upcoming work and TODO items.
-Please get involved by working on one of these items or by suggesting something else!
-Feel free to grab one of the straightforward [#mgmtlove](https://github.com/purpleidea/mgmt/labels/mgmtlove) issues if you're a first time contributor to the project or if you're unsure about what to hack on!
+Please get involved by working on one of these items or by suggesting something
+else!
 
 ## Bugs:
-Please set the `DEBUG` constant in [main.go](https://github.com/purpleidea/mgmt/blob/master/main.go) to `true`, and post the logs when you report the [issue](https://github.com/purpleidea/mgmt/issues).
-Bonus points if you provide a [shell](https://github.com/purpleidea/mgmt/tree/master/test/shell) or [OMV](https://github.com/purpleidea/mgmt/tree/master/test/omv) reproducible test case.
-Feel free to read my article on [debugging golang programs](https://ttboj.wordpress.com/2016/02/15/debugging-golang-programs/).
+
+Please set the `DEBUG` constant in [main.go](https://github.com/purpleidea/mgmt/blob/master/main.go)
+to `true`, and post the logs when you report the [issue](https://github.com/purpleidea/mgmt/issues).
+Bonus points if you provide a [shell](https://github.com/purpleidea/mgmt/tree/master/test/shell)
+or [OMV](https://github.com/purpleidea/mgmt/tree/master/test/omv) reproducible
+test case.
+Feel free to read my article on [debugging golang programs](https://purpleidea.com/blog/2016/02/15/debugging-golang-programs/).
 
 ## Patches:
+
 We'd love to have your patches! Please send them by email, or as a pull request.
 
 ## On the web:
-| Author | Format | Subject |
-|---|---|---|
-| James Shubin | blog | [Next generation configuration mgmt](https://ttboj.wordpress.com/2016/01/18/next-generation-configuration-mgmt/) |
-| James Shubin | video | [Introductory recording from DevConf.cz 2016](https://www.youtube.com/watch?v=GVhpPF0j-iE&html5=1) |
-| James Shubin | video | [Introductory recording from CfgMgmtCamp.eu 2016](https://www.youtube.com/watch?v=fNeooSiIRnA&html5=1) |
-| Julian Dunn | video | [On mgmt at CfgMgmtCamp.eu 2016](https://www.youtube.com/watch?v=kfF9IATUask&t=1949&html5=1) |
-| Walter Heck | slides | [On mgmt at CfgMgmtCamp.eu 2016](http://www.slideshare.net/olindata/configuration-management-time-for-a-4th-generation/3) |
-| Marco Marongiu | blog | [On mgmt](http://syslog.me/2016/02/15/leap-or-die/) |
-| Felix Frank | blog | [From Catalog To Mgmt (on puppet to mgmt "transpiling")](https://ffrank.github.io/features/2016/02/18/from-catalog-to-mgmt/) |
-| James Shubin | blog | [Automatic edges in mgmt (...and the pkg resource)](https://ttboj.wordpress.com/2016/03/14/automatic-edges-in-mgmt/) |
-| James Shubin | blog | [Automatic grouping in mgmt](https://ttboj.wordpress.com/2016/03/30/automatic-grouping-in-mgmt/) |
-| John Arundel | tweet | [“Puppet’s days are numbered.”](https://twitter.com/bitfield/status/732157519142002688) |
-| Felix Frank | blog | [Puppet, Meet Mgmt (on puppet to mgmt internals)](https://ffrank.github.io/features/2016/06/12/puppet,-meet-mgmt/) |
-| Felix Frank | blog | [Puppet Powered Mgmt (puppet to mgmt tl;dr)](https://ffrank.github.io/features/2016/06/19/puppet-powered-mgmt/) |
-| James Shubin | blog | [Automatic clustering in mgmt](https://ttboj.wordpress.com/2016/06/20/automatic-clustering-in-mgmt/) |
-| James Shubin | video | [Recording from CoreOSFest 2016](https://www.youtube.com/watch?v=KVmDCUA42wc&html5=1) |
-| James Shubin | video | [Recording from DebConf16](http://meetings-archive.debian.net/pub/debian-meetings/2016/debconf16/Next_Generation_Config_Mgmt.webm) ([Slides](https://annex.debconf.org//debconf-share/debconf16/slides/15-next-generation-config-mgmt.pdf)) |
-| Felix Frank | blog | [Edging It All In (puppet and mgmt edges)](https://ffrank.github.io/features/2016/07/12/edging-it-all-in/) |
-| Felix Frank | blog | [Translating All The Things (puppet to mgmt translation warnings)](https://ffrank.github.io/features/2016/08/19/translating-all-the-things/) |
-| James Shubin | video | [Recording from systemd.conf 2016](https://www.youtube.com/watch?v=jB992Zb3nH0&html5=1) |
-| James Shubin | blog | [Remote execution in mgmt](https://ttboj.wordpress.com/2016/10/07/remote-execution-in-mgmt/) |
-| James Shubin | video | [Recording from High Load Strategy 2016](https://vimeo.com/191493409) |
-| James Shubin | video | [Recording from NLUUG 2016](https://www.youtube.com/watch?v=MmpwOQAb_SE&html5=1) |
-| James Shubin | blog | [Send/Recv in mgmt](https://ttboj.wordpress.com/2016/12/07/sendrecv-in-mgmt/) |
-| James Shubin | blog | [Metaparameters in mgmt](https://ttboj.wordpress.com/2017/03/01/metaparameters-in-mgmt/) |
-| James Shubin | video | [Recording from Incontro DevOps 2017](https://vimeo.com/212241877) |
-| Yves Brissaud | blog | [mgmt aux HumanTalks Grenoble (french)](http://log.winsos.net/2017/04/12/mgmt-aux-human-talks-grenoble.html) |
-| James Shubin | video | [Recording from OSDC Berlin 2017](https://www.youtube.com/watch?v=LkEtBVLfygE&html5=1) |
 
-##
+[Read what people are saying and publishing about mgmt!](docs/on-the-web.md)
 
 Happy hacking!

TODO.md

@@ -1,4 +1,5 @@
 # TODO
+
 If you're looking for something to do, look here!
 Let us know if you're working on one of the items.
 If you'd like something to work on, ping @purpleidea and I'll create an issue
@@ -6,63 +7,78 @@ tailored especially for you! Just let me know your approximate golang skill
 level and how many hours you'd like to spend on the patch.
 
 ## Package resource
+
 - [ ] getfiles support on debian [bug](https://github.com/hughsie/PackageKit/issues/118)
 - [ ] directory info on fedora [bug](https://github.com/hughsie/PackageKit/issues/117)
 - [ ] dnf blocker [bug](https://github.com/hughsie/PackageKit/issues/110)
 
 ## File resource [bug](https://github.com/purpleidea/mgmt/issues/64) [:heart:](https://github.com/purpleidea/mgmt/labels/mgmtlove)
+
 - [ ] recurse limit support [:heart:](https://github.com/purpleidea/mgmt/labels/mgmtlove)
 - [ ] fanotify support [bug](https://github.com/go-fsnotify/fsnotify/issues/114)
 
 ## Svc resource
+
 - [ ] base resource improvements
 
 ## Exec resource
+
 - [ ] base resource improvements
 
 ## Timer resource
+
 - [ ] increment algorithm (linear, exponential, etc...) [:heart:](https://github.com/purpleidea/mgmt/labels/mgmtlove)
 
 ## User/Group resource
-- [ ] base resource [:heart:](https://github.com/purpleidea/mgmt/labels/mgmtlove)
+
 - [ ] automatic edges to file resource [:heart:](https://github.com/purpleidea/mgmt/labels/mgmtlove)
 
 ## Virt (libvirt) resource
+
 - [ ] base resource improvements [:heart:](https://github.com/purpleidea/mgmt/labels/mgmtlove)
 
 ## Net (systemd-networkd) resource
+
 - [ ] base resource
 
 ## Nspawn (systemd-nspawn) resource
+
 - [ ] base resource [:heart:](https://github.com/purpleidea/mgmt/labels/mgmtlove)
 
 ## Mount (systemd-mount) resource
+
 - [ ] base resource [:heart:](https://github.com/purpleidea/mgmt/labels/mgmtlove)
 
 ## Cron (systemd-timer) resource
+
 - [ ] base resource [:heart:](https://github.com/purpleidea/mgmt/labels/mgmtlove)
 
 ## Http resource
+
 - [ ] base resource [:heart:](https://github.com/purpleidea/mgmt/labels/mgmtlove)
 
 ## Etcd improvements
+
 - [ ] fix embedded etcd master race
 
 ## Torrent/dht file transfer
+
 - [ ] base plumbing
 
 ## GPG/Auth improvements
+
 - [ ] base plumbing
 
 ## Language improvements
-- [ ] language design
-- [ ] lexer/parser
+
+- [ ] more core functions
 - [ ] automatic language formatter, ala `gofmt`
 - [ ] gedit/gnome-builder/gtksourceview syntax highlighting
 - [ ] vim syntax highlighting
-- [ ] emacs syntax highlighting
+- [x] emacs syntax highlighting: see `misc/emacs/`
 
 ## Other
+
 - [ ] better error/retry handling
 - [ ] deb package target in Makefile
 - [ ] reproducible builds

bindata/Makefile

@@ -0,0 +1,38 @@
+# Mgmt
+# Copyright (C) 2013-2018+ James Shubin and the project contributors
+# Written by James Shubin <james@shubin.ca> and the project contributors
+#
+# This program is free software: you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation, either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+# The bindata target generates go files from any source defined below. To use
+# the files, import the "bindata" package and use:
+# `bytes, err := bindata.Asset("FILEPATH")`
+# where FILEPATH is the path of the original input file relative to `bindata/`.
+
+.PHONY: build clean
+default: build
+
+build: bindata.go
+
+# add more input files as dependencies at the end here...
+bindata.go: ../COPYING
+	# go-bindata --pkg bindata -o <OUTPUT> <INPUT>
+	go-bindata --pkg bindata -o ./$@ $^
+	# gofmt the output file
+	gofmt -s -w $@
+	@ROOT=$$(dirname "$${BASH_SOURCE}")/.. && $$ROOT/misc/header.sh '$@'
+
+clean:
+	# remove generated bindata/*.go
+	@ROOT=$$(dirname "$${BASH_SOURCE}")/.. && rm -f *.go

converger/converger.go

@@ -1,5 +1,5 @@
 // Mgmt
-// Copyright (C) 2013-2017+ James Shubin and the project contributors
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
 // Written by James Shubin <james@shubin.ca> and the project contributors
 //
 // This program is free software: you can redistribute it and/or modify

debian/.gitignore

@@ -0,0 +1,7 @@
+*.debhelper.log
+*debhelper
+changelog
+debhelper-build-stamp
+files
+mgmt.substvars
+mgmt/*

debian/compat

@@ -0,0 +1 @@
+9

debian/control

@@ -0,0 +1,17 @@
+Source: mgmt
+Maintainer: Johan Bloemberg (aequitas) <mgmt@ijohan.nl>
+Build-Depends:
+               debhelper,
+               devscripts,
+               dh-golang,
+               dh-systemd,
+               golang-go,
+
+Package: mgmt
+Architecture: any
+Depends: ${shlibs:Depends}, ${misc:Depends}, packagekit
+Suggests:      graphviz
+Description: mgmt: next generation config management!
+ The mgmt tool is a next generation config management prototype. It's
+ not yet ready for production, but we hope to get there soon. Get
+ involved today!

debian/copyright

@@ -0,0 +1,21 @@
+Format: http://www.debian.org/doc/packaging-manuals/copyright-format/1.0/
+Upstream-Name: mgmt
+Source: <https://github.com/purpleidea/mgmt>
+
+Files: *
+Copyright: Copyright (C) 2013-2018+ James Shubin and the project contributors
+License: GPL-3.0
+
+License: GPL-3.0
+  This program is free software: you can redistribute it and/or modify
+  it under the terms of the GNU General Public License as published by
+  the Free Software Foundation, either version 3 of the License, or
+  (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program.  If not, see <https://www.gnu.org/licenses/>.

debian/mgmt.docs

@@ -0,0 +1,11 @@
+AUTHORS
+COPYING
+COPYRIGHT
+README.md
+THANKS
+TODO.md
+docs
+examples
+misc/bashrc.sh
+misc/delta-cpu.sh
+misc/mgmt.service

debian/mgmt.install

@@ -0,0 +1,2 @@
+mgmt usr/bin
+misc/mgmt.service /lib/systemd/system

debian/rules

@@ -0,0 +1,15 @@
+#!/usr/bin/make -f
+
+export DH_OPTIONS
+export DH_GOPKG := mgmt
+export DH_GOLANG_INSTALL_ALL := 1
+unexport GOROOT
+
+override_dh_auto_build:
+	make build
+
+override_dh_auto_test:
+	@echo "Tests are disabled for now"
+
+%:
+	dh $@ --with=systemd

doc.go

@@ -1,5 +1,5 @@
 // Mgmt
-// Copyright (C) 2013-2017+ James Shubin and the project contributors
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
 // Written by James Shubin <james@shubin.ca> and the project contributors
 //
 // This program is free software: you can redistribute it and/or modify

docker/Dockerfile

@@ -1,10 +1,10 @@
-FROM golang:1.6.2
+FROM golang:1.9
 
 MAINTAINER Michał Czeraszkiewicz <contact@czerasz.com>
 
 # Set the reset cache variable
 # Read more here: http://czerasz.com/2014/11/13/docker-tip-and-tricks/#use-refreshedat-variable-for-better-cache-control
-ENV REFRESHED_AT 2016-05-10
+ENV REFRESHED_AT 2017-11-16
 
 # Update the package list to be able to use required packages
 RUN apt-get update

docker/Dockerfile.build

@@ -0,0 +1,12 @@
+FROM centos:7
+MAINTAINER Karim Boumedhel <karimboumedhel@gmail.com>
+
+ENV GOPATH=/root/gopath
+ENV PATH=/opt/rh/rh-ruby22/root/usr/bin:/root/gopath/bin:/usr/local/sbin:/sbin:/bin:/usr/sbin:/usr/bin:/usr/local/go/bin
+ENV LD_LIBRARY_PATH=/opt/rh/rh-ruby22/root/usr/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
+ENV PKG_CONFIG_PATH=/opt/rh/rh-ruby22/root/usr/lib64/pkgconfig${PKG_CONFIG_PATH:+:${PKG_CONFIG_PATH}}
+
+RUN yum -y install epel-release wget unzip git make which centos-release-scl gcc && sed -i "s/enabled=0/enabled=1/" /etc/yum.repos.d/epel-testing.repo && yum -y install rh-ruby22 && wget -O /opt/go1.9.1.linux-amd64.tar.gz https://storage.googleapis.com/golang/go1.9.1.linux-amd64.tar.gz && tar -C /usr/local -xzf /opt/go1.9.1.linux-amd64.tar.gz
+RUN mkdir -p $GOPATH/src/github.com/purpleidea && cd $GOPATH/src/github.com/purpleidea && git clone --recursive https://github.com/purpleidea/mgmt
+RUN go get -u gopkg.in/alecthomas/gometalinter.v1 && cd $GOPATH/src/github.com/purpleidea/mgmt && make deps && make build
+CMD ["/bin/bash"]

docker/Dockerfile.development

@@ -1,10 +1,10 @@
-FROM golang:1.6.2
+FROM golang:1.9
 
 MAINTAINER Michał Czeraszkiewicz <contact@czerasz.com>
 
 # Set the reset cache variable
 # Read more here: http://czerasz.com/2014/11/13/docker-tip-and-tricks/#use-refreshedat-variable-for-better-cache-control
-ENV REFRESHED_AT 2016-05-14
+ENV REFRESHED_AT 2017-11-16
 
 RUN apt-get update
 
@@ -27,5 +27,8 @@ WORKDIR /home/$USER_NAME/mgmt
 # Install dependencies
 RUN make deps
 
+# Chown $GOPATH
+RUN chown -R ${USER_ID}:${GROUP_ID} /go
+
 # Change user
 USER ${USER_NAME}

docker/Dockerfile.static

@@ -0,0 +1,9 @@
+FROM centos:7
+MAINTAINER Karim Boumedhel <karimboumedhel@gmail.com>
+
+RUN yum -y install augeas-libs libvirt-libs && yum clean all
+ADD mgmt /usr/bin
+RUN chmod 700 /usr/bin/mgmt
+
+ENTRYPOINT ["/usr/bin/mgmt"]
+CMD ["-h"]

docker/scripts/exec-development

@@ -0,0 +1,18 @@
+#!/bin/bash
+
+# runs command provided as argument inside a development (Linux) Docker container
+
+# Stop on any error
+set -e
+
+script_directory="$( cd "$( dirname "$0" )" && pwd )"
+project_directory=$script_directory/../..
+
+# Specify the Docker image name
+image_name='purpleidea/mgmt:development'
+
+# Run container in development mode
+docker run --rm --name=mgm_development --user=mgmt \
+	-v "$project_directory:/go/src/github.com/purpleidea/mgmt/" \
+	-w /go/src/github.com/purpleidea/mgmt/ \
+	-it "$image_name" /bin/bash -c "$*"

docs/conf.py

@@ -51,7 +51,7 @@ master_doc = 'index'
 
 # General information about the project.
 project = u'mgmt'
-copyright = u'2013-2017+ James Shubin and the project contributors'
+copyright = u'2013-2018+ James Shubin and the project contributors'
 author = u'James Shubin'
 
 # The version info for the project you're documenting, acts as replacement for

docs/development.md

@@ -0,0 +1,49 @@
+# Development
+
+This document contains some additional information and help regarding
+developing `mgmt`. Useful tools, conventions, etc.
+
+Be sure to read [quick start guide](docs/quick-start-guide.md) first.
+
+## Testing
+
+This project has both unit tests in the form of golang tests and integration
+tests using shell scripting.
+
+Native golang tests are preferred over tests written in our shell testing
+framework. Please see [https://golang.org/pkg/testing/](https://golang.org/pkg/testing/)
+for more information.
+
+To run all tests:
+
+```
+make test
+```
+
+There is a library of quick and small integration tests for the language and
+YAML related things, check out [`test/shell/`](/test/shell). Adding a test is as
+easy as copying one of the files in [`test/shell/`](/test/shell) and adapting
+it.
+
+This test suite won't run by default (unless when on CI server) and needs to be
+called explictly using:
+
+```
+make test-shell
+```
+
+Or run an individual shell test using:
+
+```
+make test-shell-load0
+```
+
+Tip: you can use TAB completion with `make` to quickly get a list of possible
+individual tests to run.
+
+## Tools, integrations, IDE's etc
+
+### IDE/Editor support
+
+- Emacs: see `misc/emacs/`
+- [Textmate](https://github.com/aequitas/mgmt.tmbundle)

docs/documentation.md

@@ -1,9 +1,4 @@
-# mgmt
-
-Available from:
-[https://github.com/purpleidea/mgmt/](https://github.com/purpleidea/mgmt/)
-
-This documentation is available in: [Markdown](https://github.com/purpleidea/mgmt/blob/master/docs/documentation.md) or [PDF](https://pdfdoc-purpleidea.rhcloud.com/pdf/https://github.com/purpleidea/mgmt/blob/master/docs/documentation.md) format.
+# General documentation
 
 ## Overview
 
@@ -18,24 +13,21 @@ foundation in and for, new and existing software.
 
 For more information, you may like to read some blog posts from the author:
 
-* [Next generation config mgmt](https://ttboj.wordpress.com/2016/01/18/next-generation-configuration-mgmt/)
-* [Automatic edges in mgmt](https://ttboj.wordpress.com/2016/03/14/automatic-edges-in-mgmt/)
-* [Automatic grouping in mgmt](https://ttboj.wordpress.com/2016/03/30/automatic-grouping-in-mgmt/)
-* [Automatic clustering in mgmt](https://ttboj.wordpress.com/2016/06/20/automatic-clustering-in-mgmt/)
-* [Remote execution in mgmt](https://ttboj.wordpress.com/2016/10/07/remote-execution-in-mgmt/)
-* [Send/Recv in mgmt](https://ttboj.wordpress.com/2016/12/07/sendrecv-in-mgmt/)
-* [Metaparameters in mgmt](https://ttboj.wordpress.com/2017/03/01/metaparameters-in-mgmt/)
+* [Next generation config mgmt](https://purpleidea.com/blog/2016/01/18/next-generation-configuration-mgmt/)
+* [Automatic edges in mgmt](https://purpleidea.com/blog/2016/03/14/automatic-edges-in-mgmt/)
+* [Automatic grouping in mgmt](https://purpleidea.com/blog/2016/03/30/automatic-grouping-in-mgmt/)
+* [Automatic clustering in mgmt](https://purpleidea.com/blog/2016/06/20/automatic-clustering-in-mgmt/)
+* [Remote execution in mgmt](https://purpleidea.com/blog/2016/10/07/remote-execution-in-mgmt/)
+* [Send/Recv in mgmt](https://purpleidea.com/blog/2016/12/07/sendrecv-in-mgmt/)
+* [Metaparameters in mgmt](https://purpleidea.com/blog/2017/03/01/metaparameters-in-mgmt/)
 
-There is also an [introductory video](http://meetings-archive.debian.net/pub/debian-meetings/2016/debconf16/Next_Generation_Config_Mgmt.webm) available.
-Older videos and other material [is available](https://github.com/purpleidea/mgmt/#on-the-web).
+There is also an [introductory video](https://www.youtube.com/watch?v=LkEtBVLfygE&html5=1)
+available. Older videos and other material [is available](on-the-web.md).
 
 ## Setup
 
-During this prototype phase, the tool can be run out of the source directory.
-You'll probably want to use ```./run.sh run --yaml examples/graph1.yaml``` to
-get started. Beware that this _can_ cause data loss. Understand what you're
-doing first, or perform these actions in a virtual environment such as the one
-provided by [Oh-My-Vagrant](https://github.com/purpleidea/oh-my-vagrant).
+You'll probably want to read the [quick start guide](quick-start-guide.md) to
+get going.
 
 ## Features
 
@@ -71,7 +63,7 @@ the meta attributes of that resource to `false`.
 #### Blog post
 
 You can read the introductory blog post about this topic here:
-[https://ttboj.wordpress.com/2016/03/14/automatic-edges-in-mgmt/](https://ttboj.wordpress.com/2016/03/14/automatic-edges-in-mgmt/)
+[https://purpleidea.com/blog/2016/03/14/automatic-edges-in-mgmt/](https://purpleidea.com/blog/2016/03/14/automatic-edges-in-mgmt/)
 
 ### Autogrouping
 
@@ -90,7 +82,7 @@ the meta attributes of that resource to `false`.
 #### Blog post
 
 You can read the introductory blog post about this topic here:
-[https://ttboj.wordpress.com/2016/03/30/automatic-grouping-in-mgmt/](https://ttboj.wordpress.com/2016/03/30/automatic-grouping-in-mgmt/)
+[https://purpleidea.com/blog/2016/03/30/automatic-grouping-in-mgmt/](https://purpleidea.com/blog/2016/03/30/automatic-grouping-in-mgmt/)
 
 ### Automatic clustering
 
@@ -106,7 +98,7 @@ with the `--seeds` variable.
 #### Blog post
 
 You can read the introductory blog post about this topic here:
-[https://ttboj.wordpress.com/2016/06/20/automatic-clustering-in-mgmt/](https://ttboj.wordpress.com/2016/06/20/automatic-clustering-in-mgmt/)
+[https://purpleidea.com/blog/2016/06/20/automatic-clustering-in-mgmt/](https://purpleidea.com/blog/2016/06/20/automatic-clustering-in-mgmt/)
 
 ### Remote ("agent-less") mode
 
@@ -133,7 +125,7 @@ which need to exchange information that is only available at run time.
 #### Blog post
 
 You can read the introductory blog post about this topic here:
-[https://ttboj.wordpress.com/2016/10/07/remote-execution-in-mgmt/](https://ttboj.wordpress.com/2016/10/07/remote-execution-in-mgmt/)
+[https://purpleidea.com/blog/2016/10/07/remote-execution-in-mgmt/](https://purpleidea.com/blog/2016/10/07/remote-execution-in-mgmt/)
 
 ### Puppet support
 
@@ -145,15 +137,15 @@ Invoke `mgmt` with the `--puppet` switch, which supports 3 variants:
 
 1. Request the configuration from the Puppet Master (like `puppet agent` does)
 
-        mgmt run --puppet agent
+	`mgmt run --puppet agent`
 
 2. Compile a local manifest file (like `puppet apply`)
 
-        mgmt run --puppet /path/to/my/manifest.pp
+	`mgmt run --puppet /path/to/my/manifest.pp`
 
 3. Compile an ad hoc manifest from the commandline (like `puppet apply -e`)
 
-        mgmt run --puppet 'file { "/etc/ntp.conf": ensure => file }'
+	`mgmt run --puppet 'file { "/etc/ntp.conf": ensure => file }'`
 
 For more details and caveats see [Puppet.md](Puppet.md).
 
@@ -162,283 +154,41 @@ For more details and caveats see [Puppet.md](Puppet.md).
 An introductory post on the Puppet support is on
 [Felix's blog](http://ffrank.github.io/features/2016/06/19/puppet-powered-mgmt/).
 
-## Resources
-
-This section lists all the built-in resources and their properties. The
-resource primitives in `mgmt` are typically more powerful than resources in
-other configuration management systems because they can be event based which
-lets them respond in real-time to converge to the desired state. This property
-allows you to build more complex resources that you probably hadn't considered
-in the past.
-
-In addition to the resource specific properties, there are resource properties
-(otherwise known as parameters) which can apply to every resource. These are
-called [meta parameters](#meta-parameters) and are listed separately. Certain
-meta parameters aren't very useful when combined with certain resources, but
-in general, it should be fairly obvious, such as when combining the `noop` meta
-parameter with the [Noop](#Noop) resource.
-
-* [Augeas](#Augeas): Manipulate files using augeas.
-* [Exec](#Exec): Execute shell commands on the system.
-* [File](#File): Manage files and directories.
-* [Hostname](#Hostname): Manages the hostname on the system.
-* [KV](#KV): Set a key value pair in our shared world database.
-* [Msg](#Msg): Send log messages.
-* [Noop](#Noop): A simple resource that does nothing.
-* [Nspawn](#Nspawn): Manage systemd-machined nspawn containers.
-* [Password](#Password): Create random password strings.
-* [Pkg](#Pkg):  Manage system packages with PackageKit.
-* [Svc](#Svc): Manage system systemd services.
-* [Timer](#Timer): Manage system systemd services.
-* [Virt](#Virt): Manage virtual machines with libvirt.
-
-
-### Augeas
-
-The augeas resource uses [augeas](http://augeas.net/) commands to manipulate
-files.
-
-### Exec
-
-The exec resource can execute commands on your system.
-
-### File
-
-The file resource manages files and directories. In `mgmt`, directories are
-identified by a trailing slash in their path name. File have no such slash.
-
-It has the following properties:
-
-- `path`: file path (directories have a trailing slash here)
-- `content`: raw file content
-- `state`: either `exists` (the default value) or `absent`
-- `mode`: octal unix file permissions
-- `owner`: username or uid for the file owner
-- `group`: group name or gid for the file group
-
-#### Path
-
-The path property specifies the file or directory that we are managing.
-
-#### Content
-
-The content property is a string that specifies the desired file contents.
-
-#### Source
-
-The source property points to a source file or directory path that we wish to
-copy over and use as the desired contents for our resource.
-
-#### State
-
-The state property describes the action we'd like to apply for the resource. The
-possible values are: `exists` and `absent`.
-
-#### Recurse
-
-The recurse property limits whether file resource operations should recurse into
-and monitor directory contents with a depth greater than one.
-
-#### Force
-
-The force property is required if we want the file resource to be able to change
-a file into a directory or vice-versa. If such a change is needed, but the force
-property is not set to `true`, then this file resource will error.
-
-### Hostname
-
-The hostname resource manages static, transient/dynamic and pretty hostnames
-on the system and watches them for changes.
-
-#### static_hostname
-The static hostname is the one configured in /etc/hostname or a similar
-file.
-It is chosen by the local user. It is not always in sync with the current
-host name as returned by the gethostname() system call.
-
-#### transient_hostname
-The transient / dynamic hostname is the one configured via the kernel's
-sethostbyname().
-It can be different from the static hostname in case DHCP or mDNS have been
-configured to change the name based on network information.
-
-#### pretty_hostname
-The pretty hostname is a free-form UTF8 host name for presentation to the user.
-
-#### hostname
-Hostname is the fallback value for all 3 fields above, if only `hostname` is
-specified, it will set all 3 fields to this value.
-
-### KV
-
-The KV resource sets a key and value pair in the global world database. This is
-quite useful for setting a flag after a number of resources have run. It will
-ignore database updates to the value that are greater in compare order than the
-requested key if the `SkipLessThan` parameter is set to true. If we receive a
-refresh, then the stored value will be reset to the requested value even if the
-stored value is greater.
-
-#### Key
-The string key used to store the key.
-
-#### Value
-The string value to set. This can also be set via Send/Recv.
-
-#### SkipLessThan
-If this parameter is set to `true`, then it will ignore updating the value as
-long as the database versions are greater than the requested value. The compare
-operation used is based on the `SkipCmpStyle` parameter.
-
-#### SkipCmpStyle
-By default this converts the string values to integers and compares them as you
-would expect.
-
-### Msg
-
-The msg resource sends messages to the main log, or an external service such
-as systemd's journal.
-
-### Noop
-
-The noop resource does absolutely nothing. It does have some utility in testing
-`mgmt` and also as a placeholder in the resource graph.
-
-### Nspawn
-
-The nspawn resource is used to manage systemd-machined style containers.
-
-### Password
-
-The password resource can generate a random string to be used as a password. It
-will re-generate the password if it receives a refresh notification.
-
-### Pkg
-
-The pkg resource is used to manage system packages. This resource works on many
-different distributions because it uses the underlying packagekit facility which
-supports different backends for different environments. This ensures that we
-have great Debian (deb/dpkg) and Fedora (rpm/dnf) support simultaneously.
-
-### Svc
-
-The service resource is still very WIP. Please help us my improving it!
-
-### Timer
-
-This resource needs better documentation. Please help us my improving it!
-
-### Virt
-
-The virt resource can manage virtual machines via libvirt.
-
-## Usage and frequently asked questions
-(Send your questions as a patch to this FAQ! I'll review it, merge it, and
-respond by commit with the answer.)
-
-### Why did you start this project?
-
-I wanted a next generation config management solution that didn't have all of
-the design flaws or limitations that the current generation of tools do, and no
-tool existed!
-
-### Why did you use etcd? What about consul?
-
-Etcd and consul are both written in golang, which made them the top two
-contenders for my prototype. Ultimately a choice had to be made, and etcd was
-chosen, but it was also somewhat arbitrary. If there is available interest,
-good reasoning, *and* patches, then we would consider either switching or
-supporting both, but this is not a high priority at this time.
-
-### Can I use an existing etcd cluster instead of the automatic embedded servers?
-
-Yes, it's possible to use an existing etcd cluster instead of the automatic,
-elastic embedded etcd servers. To do so, simply point to the cluster with the
-`--seeds` variable, the same way you would if you were seeding a new member to
-an existing mgmt cluster.
-
-The downside to this approach is that you won't benefit from the automatic
-elastic nature of the embedded etcd servers, and that you're responsible if you
-accidentally break your etcd cluster, or if you use an unsupported version.
-
-### What does the error message about an inconsistent dataDir mean?
-
-If you get an error message similar to:
-
-```
-Etcd: Connect: CtxError...
-Etcd: CtxError: Reason: CtxDelayErr(5s): No endpoints available yet!
-Etcd: Connect: Endpoints: []
-Etcd: The dataDir (/var/lib/mgmt/etcd) might be inconsistent or corrupt.
-```
-
-This happens when there are a series of fatal connect errors in a row. This can
-happen when you start `mgmt` using a dataDir that doesn't correspond to the
-current cluster view. As a result, the embedded etcd server never finishes
-starting up, and as a result, a default endpoint never gets added. The solution
-is to either reconcile the mistake, and if there is no important data saved, you
-can remove the etcd dataDir. This is typically `/var/lib/mgmt/etcd/member/`.
-
-### Why do resources have both a `Compare` method and an `IFF` (on the UID) method?
-
-The `Compare()` methods are for determining if two resources are effectively the
-same, which is used to make graph change delta's efficient. This is when we want
-to change from the current running graph to a new graph, but preserve the common
-vertices. Since we want to make this process efficient, we only update the parts
-that are different, and leave everything else alone. This `Compare()` method can
-tell us if two resources are the same.
-
-The `IFF()` method is part of the whole UID system, which is for discerning if a
-resource meets the requirements another expects for an automatic edge. This is
-because the automatic edge system assumes a unified UID pattern to test for
-equality. In the future it might be helpful or sane to merge the two similar
-comparison functions although for now they are separate because they are
-actually answer different questions.
-
-### Did you know that there is a band named `MGMT`?
-
-I didn't realize this when naming the project, and it is accidental. After much
-anguishing, I chose the name because it was short and I thought it was
-appropriately descriptive. If you need a less ambiguous search term or phrase,
-you can try using `mgmtconfig` or `mgmt config`.
-
-### You didn't answer my question, or I have a question!
-
-It's best to ask on [IRC](https://webchat.freenode.net/?channels=#mgmtconfig)
-to see if someone can help you. Once we get a big enough community going, we'll
-add a mailing list. If you don't get any response from the above, you can
-contact me through my [technical blog](https://ttboj.wordpress.com/contact/)
-and I'll do my best to help. If you have a good question, please add it as a
-patch to this documentation. I'll merge your question, and add a patch with the
-answer!
-
 ## Reference
+
 Please note that there are a number of undocumented options. For more
 information on these options, please view the source at:
 [https://github.com/purpleidea/mgmt/](https://github.com/purpleidea/mgmt/).
 If you feel that a well used option needs documenting here, please patch it!
 
 ### Overview of reference
+
 * [Meta parameters](#meta-parameters): List of available resource meta parameters.
 * [Graph definition file](#graph-definition-file): Main graph definition file.
 * [Command line](#command-line): Command line parameters.
 * [Compilation options](#compilation-options): Compilation options.
 
 ### Meta parameters
+
 These meta parameters are special parameters (or properties) which can apply to
 any resource. The usefulness of doing so will depend on the particular meta
 parameter and resource combination.
 
 #### AutoEdge
+
 Boolean. Should we generate auto edges for this resource?
 
 #### AutoGroup
+
 Boolean. Should we attempt to automatically group this resource with others?
 
 #### Noop
+
 Boolean. Should the Apply portion of the CheckApply method of the resource
 make any changes? Noop is a concatenation of no-operation.
 
 #### Retry
+
 Integer. The number of times to retry running the resource on error. Use -1 for
 infinite. This currently applies for both the Watch operation (which can fail)
 and for the CheckApply operation. While they could have separate values, I've
@@ -446,6 +196,7 @@ decided to use the same ones for both until there's a proper reason to want to
 do something differently for the Watch errors.
 
 #### Delay
+
 Integer. Number of milliseconds to wait between retries. The same value is
 shared between the Watch and CheckApply retries. This currently applies for both
 the Watch operation (which can fail) and for the CheckApply operation. While
@@ -454,6 +205,7 @@ until there's a proper reason to want to do something differently for the Watch
 errors.
 
 #### Poll
+
 Integer. Number of seconds to wait between `CheckApply` checks. If this is
 greater than zero, then the standard event based `Watch` mechanism for this
 resource is replaced with a simple polling mechanism. In general, this is not
@@ -471,6 +223,7 @@ which is another way of saying that if the resource finally settles down to give
 the graph enough time, it can probably converge.
 
 #### Limit
+
 Float. Maximum rate of `CheckApply` runs started per second. Useful to limit
 an especially _eventful_ process from causing excessive checks to run. This
 defaults to `+Infinity` which adds no limiting. If you change this value, you
@@ -478,12 +231,14 @@ will also need to change the `Burst` value to a non-zero value. Please see the
 [rate](https://godoc.org/golang.org/x/time/rate) package for more information.
 
 #### Burst
+
 Integer. Burst is the maximum number of runs which can happen without invoking
 the rate limiter as designated by the `Limit` value. If the `Limit` is not set
 to `+Infinity`, this must be a non-zero value. Please see the
 [rate](https://godoc.org/golang.org/x/time/rate) package for more information.
 
 #### Sema
+
 List of string ids. Sema is a P/V style counting semaphore which can be used to
 limit parallelism during the CheckApply phase of resource execution. Each
 resource can have `N` different semaphores which share a graph global namespace.
@@ -495,30 +250,37 @@ id's include: `some_id`, `hello:42`, `not:smart:4` and `:13`. It is expected
 that the last bare example be only used by the engine to add a global semaphore.
 
 ### Graph definition file
+
 graph.yaml is the compiled graph definition file. The format is currently
 undocumented, but by looking through the [examples/](https://github.com/purpleidea/mgmt/tree/master/examples)
 you can probably figure out most of it, as it's fairly intuitive.
 
 ### Command line
+
 The main interface to the `mgmt` tool is the command line. For the most recent
 documentation, please run `mgmt --help`.
 
 #### `--yaml <graph.yaml>`
+
 Point to a graph file to run.
 
 #### `--converged-timeout <seconds>`
+
 Exit if the machine has converged for approximately this many seconds.
 
 #### `--max-runtime <seconds>`
+
 Exit when the agent has run for approximately this many seconds. This is not
 generally recommended, but may be useful for users who know what they're doing.
 
 #### `--noop`
+
 Globally force all resources into no-op mode. This also disables the export to
 etcd functionality, but does not disable resource collection, however all
 resources that are collected will have their individual noop settings set.
 
 #### `--sema <size>`
+
 Globally add a counting semaphore of this size to each resource in the graph.
 The semaphore will get given an id of `:size`. In other words if you specify a
 size of 42, you can expect a semaphore if named: `:42`. It is expected that
@@ -528,38 +290,46 @@ than zero at this time. The traditional non-parallel execution found in config
 management tools such as `Puppet` can be obtained with `--sema 1`.
 
 #### `--remote <graph.yaml>`
+
 Point to a graph file to run on the remote host specified within. This parameter
 can be used multiple times if you'd like to remotely run on multiple hosts in
 parallel.
 
 #### `--allow-interactive`
+
 Allow interactive prompting for SSH passwords if there is no authentication
 method that works.
 
 #### `--ssh-priv-id-rsa`
+
 Specify the path for finding SSH keys. This defaults to `~/.ssh/id_rsa`. To
 never use this method of authentication, set this to the empty string.
 
 #### `--cconns`
+
 The maximum number of concurrent remote ssh connections to run. This defaults
 to `0`, which means unlimited.
 
 #### `--no-caching`
+
 Don't allow remote caching of the remote execution binary. This will require
 the binary to be copied over for every remote execution, but it limits the
 likelihood that there is leftover information from the configuration process.
 
 #### `--prefix <path>`
+
 Specify a path to a custom working directory prefix. This directory will get
 created if it does not exist. This usually defaults to `/var/lib/mgmt/`. This
 can't be combined with the `--tmp-prefix` option. It can be combined with the
 `--allow-tmp-prefix` option.
 
 #### `--tmp-prefix`
+
 If this option is specified, a temporary prefix will be used instead of the
 default prefix. This can't be combined with the `--prefix` option.
 
 #### `--allow-tmp-prefix`
+
 If this option is specified, we will attempt to fall back to a temporary prefix
 if the primary prefix couldn't be created. This is useful for avoiding failures
 in environments where the primary prefix may or may not be available, but you'd
@@ -596,12 +366,14 @@ GOTAGS="noaugeas novirt" make build
 ```
 
 ## Examples
-For example configurations, please consult the [examples/](https://github.com/purpleidea/mgmt/tree/master/examples) directory in the git
-source repository. It is available from:
+
+For example configurations, please consult the [examples/](https://github.com/purpleidea/mgmt/tree/master/examples)
+directory in the git source repository. It is available from:
 
 [https://github.com/purpleidea/mgmt/tree/master/examples](https://github.com/purpleidea/mgmt/tree/master/examples)
 
 ### Systemd:
+
 See [`misc/mgmt.service`](misc/mgmt.service) for a sample systemd unit file.
 This unit file is part of the RPM.
 
@@ -629,13 +401,13 @@ This is a project that I started in my free time in 2013. Development is driven
 by all of our collective patches! Dive right in, and start hacking!
 Please contact me if you'd like to invite me to speak about this at your event.
 
-You can follow along [on my technical blog](https://ttboj.wordpress.com/).
+You can follow along [on my technical blog](https://purpleidea.com/blog/).
 
 To report any bugs, please file a ticket at: [https://github.com/purpleidea/mgmt/issues](https://github.com/purpleidea/mgmt/issues).
 
 ## Authors
 
-Copyright (C) 2013-2017+ James Shubin and the project contributors
+Copyright (C) 2013-2018+ James Shubin and the project contributors
 
 Please see the
 [AUTHORS](https://github.com/purpleidea/mgmt/tree/master/AUTHORS) file
@@ -643,4 +415,4 @@ for more information.
 
 * [github](https://github.com/purpleidea/)
 * [&#64;purpleidea](https://twitter.com/#!/purpleidea)
-* [https://ttboj.wordpress.com/](https://ttboj.wordpress.com/)
+* [https://purpleidea.com/](https://purpleidea.com/)

docs/faq.md

@@ -0,0 +1,267 @@
+## Frequently asked questions
+
+(Send your questions as a patch to this FAQ! I'll review it, merge it, and
+respond by commit with the answer.)
+
+### Why did you start this project?
+
+I wanted a next generation config management solution that didn't have all of
+the design flaws or limitations that the current generation of tools do, and no
+tool existed!
+
+### How do I contribute to the project if I don't know `golang`?
+
+There are many different ways you can contribute to the project. They can be
+broadly divided into two main categories:
+
+1. With contributions written in `golang`
+2. With contributions _not_ written in `golang`
+
+If you do not know `golang`, and have no desire to learn, you can still
+contribute to mgmt by using it, testing it, writing docs, or even just by
+telling your friends about it. If you don't mind some coding, learning about the
+[mgmt language](https://purpleidea.com/blog/2018/02/05/mgmt-configuration-language/)
+might be an enjoyable experience for you. It is a small [DSL](https://en.wikipedia.org/wiki/Domain-specific_language)
+and not a general purpose programming language, and you might find it more fun
+than what you're typically used to. One of the reasons the mgmt author got into
+writing automation modules, was because he found it much more fun to build with
+a higher level DSL, than in a general purpose programming language.
+
+If you do not know `golang`, and would like to learn, are a beginner and want to
+improve your skills, or want to gain some great interdisciplinary systems
+engineering knowledge around a cool automation project, we're happy to mentor
+you. Here are some pre-requisites steps which we recommend:
+
+1. Make sure you have a somewhat recent GNU/Linux environment to hack on. A
+recent [Fedora](https://getfedora.org/) or [Debian](https://www.debian.org/)
+environment is recommended. Developing, testing, and contributing on `macOS` or
+`Windows` will be either more difficult or impossible.
+2. Ensure that you're mildly comfortable with the basics of using `git`. You can
+find a number of tutorials online.
+3. Spend between four to six hours with the [golang tour](https://tour.golang.org/).
+Skip over the longer problems, but try and get a solid overview of everything.
+If you forget something, you can always go back and repeat those parts.
+4. Connect to our [#mgmtconfig](https://webchat.freenode.net/?channels=#mgmtconfig)
+IRC channel on the [Freenode](https://freenode.net/) network. You can use any
+IRC client that you'd like, but the [hosted web portal](https://webchat.freenode.net/?channels=#mgmtconfig)
+will suffice if you don't know what else to use.
+5. Now it's time to try and starting writing a patch! We have tagged a bunch of
+[open issues as #mgmtlove](https://github.com/purpleidea/mgmt/issues?q=is%3Aissue+is%3Aopen+label%3Amgmtlove)
+for new users to have somewhere to get involved. Look through them to see if
+something interests you. If you find one, let us know you're working on it by
+leaving a comment in the ticket. We'll be around to answer questions in the IRC
+channel, and to create new issues if there wasn't something that fit your
+interests. When you submit a patch, we'll review it and give you some feedback.
+Over time, we hope you'll learn a lot while supporting the project! Now get
+hacking!
+
+### Is this project ready for production?
+
+Compared to some existing automation tools out there, mgmt is a relatively new
+project. It is probably not as feature complete as some other software, but it
+also offers a number of features which are not currently available elsewhere.
+
+Because we have not released a `1.0` release yet, we are not guaranteeing
+stability of the internal or external API's. We only change them if it's really
+necessary, and we don't expect anything particularly drastic to occur. We would
+expect it to be relatively easy to adapt your code if such changes happened.
+
+As with all software, bugs can occur, and while we make no guarantees of being
+bug-free, there are a number of things we've done to reduce the chances of one
+causing you trouble:
+
+1. Our software is written in golang, which is a memory-safe language, and which
+is known to reduce or eliminate entire classes of bugs.
+2. We have a test suite which we run on every commit, and every 24 hours. If you
+have a particular case that you'd like to test, you are welcome to add it in!
+3. The mgmt language itself offers a number of safety features. You can
+[read about them in the introductory blog post](https://purpleidea.com/blog/2018/02/05/mgmt-configuration-language/).
+
+Having said all this, as with all software, there are still missing features
+which some users might want in their production environments. We're working hard
+to get all of those implemented, but we hope that you'll get involved and help
+us finish off the ones that are most important to you. We are happy to mentor
+new contributors, and have even [tagged](https://github.com/purpleidea/mgmt/issues?q=is%3Aissue+is%3Aopen+label%3Amgmtlove)
+a number of issues if you need help getting started.
+
+Some of the current limitations include:
+
+* Auth hasn't been implemented yet, so you should only use it in trusted
+environments (not on publicly accessible networks) for now.
+* The number of built-in core functions is still small. You may encounter
+scenarios where you're missing a function. The good news is that it's relatively
+easy to add this missing functionality yourself. In time, with your help, the
+list will grow!
+* Large file distribution is not yet implemented. You might want a scenario
+where mgmt is used to distribute large files (such as `.iso` images) throughout
+your cluster. While this isn't a common use-case, it won't be possible until
+someone wants to write the patch. (Mentoring available!) You can workaround this
+easily by storing those files on a separate fileserver for the interim.
+* There isn't an ecosystem of community `modules` yet. We've got this on our
+roadmap, so please stay tuned!
+
+We hope you'll participate as an early adopter. Every additional pair of helping
+hands gets us all there faster! It's quite possible to use this to build useful
+automation today, and we hope you'll start getting familiar with the software.
+
+### Why did you use etcd? What about consul?
+
+Etcd and consul are both written in golang, which made them the top two
+contenders for my prototype. Ultimately a choice had to be made, and etcd was
+chosen, but it was also somewhat arbitrary. If there is available interest,
+good reasoning, *and* patches, then we would consider either switching or
+supporting both, but this is not a high priority at this time.
+
+### Can I use an existing etcd cluster instead of the automatic embedded servers?
+
+Yes, it's possible to use an existing etcd cluster instead of the automatic,
+elastic embedded etcd servers. To do so, simply point to the cluster with the
+`--seeds` variable, the same way you would if you were seeding a new member to
+an existing mgmt cluster.
+
+The downside to this approach is that you won't benefit from the automatic
+elastic nature of the embedded etcd servers, and that you're responsible if you
+accidentally break your etcd cluster, or if you use an unsupported version.
+
+### How can I run `mgmt` on-demand, or in `cron`, instead of continuously?
+
+By default, `mgmt` will run continuously in an attempt to keep your machine in a
+converged state, even as external forces change the current state, or as your
+time-varying desired state changes over time. (You can write code in the mgmt
+language which will let you describe a desired state which might change over
+time.)
+
+Some users might prefer to only run `mgmt` on-demand manually, or at a set
+interval via a tool like `cron`. In order to do so, `mgmt` must have a way to
+shut itself down after a single "run". This feature is possible with the
+`--converged-timeout` flag. You may specify this flag, along with a number of
+seconds as the argument, and when there has been no activity for that many
+seconds, the program will shutdown.
+
+Alternatively, while it is not recommended, if you'd like to ensure the program
+never runs for longer that a specific number of seconds, you can ask it to
+shutdown after that time interval using the `--max-runtime` flag. This also
+requires a number of seconds as an argument.
+
+#### Example:
+
+```
+./mgmt run --lang examples/lang/hello0.mcl --converged-timeout=5
+```
+
+### What does the error message about an inconsistent dataDir mean?
+
+If you get an error message similar to:
+
+```
+Etcd: Connect: CtxError...
+Etcd: CtxError: Reason: CtxDelayErr(5s): No endpoints available yet!
+Etcd: Connect: Endpoints: []
+Etcd: The dataDir (/var/lib/mgmt/etcd) might be inconsistent or corrupt.
+```
+
+This happens when there are a series of fatal connect errors in a row. This can
+happen when you start `mgmt` using a dataDir that doesn't correspond to the
+current cluster view. As a result, the embedded etcd server never finishes
+starting up, and as a result, a default endpoint never gets added. The solution
+is to either reconcile the mistake, and if there is no important data saved, you
+can remove the etcd dataDir. This is typically `/var/lib/mgmt/etcd/member/`.
+
+### Why do resources have both a `Compare` method and an `IFF` (on the UID) method?
+
+The `Compare()` methods are for determining if two resources are effectively the
+same, which is used to make graph change delta's efficient. This is when we want
+to change from the current running graph to a new graph, but preserve the common
+vertices. Since we want to make this process efficient, we only update the parts
+that are different, and leave everything else alone. This `Compare()` method can
+tell us if two resources are the same.
+
+The `IFF()` method is part of the whole UID system, which is for discerning if a
+resource meets the requirements another expects for an automatic edge. This is
+because the automatic edge system assumes a unified UID pattern to test for
+equality. In the future it might be helpful or sane to merge the two similar
+comparison functions although for now they are separate because they are
+actually answer different questions.
+
+### Does this support Windows? OSX? GNU Hurd?
+
+Mgmt probably works best on Linux, because that's what most developers use for
+serious automation workloads. Support for non-Linux operating systems isn't a
+high priority of mine, but we're happy to accept patches for missing features
+or resources that you think would make sense on your favourite platform.
+
+### Why aren't you using `glide` or `godep` for dependency management?
+
+Vendoring dependencies means that as the git master branch of each dependency
+marches on, you're left behind using an old version. As a result, bug fixes and
+improvements are not automatically brought into the project. Instead, we run our
+complete test suite against the entire project (with the latest dependencies)
+[every 24 hours](https://docs.travis-ci.com/user/cron-jobs/) to ensure that it
+all still works.
+
+Occasionally a dependency breaks API and causes a failure. In those situations,
+we're notified almost immediately, it's easy to see exactly which commit caused
+the breakage, and we can either quickly notify the author (if it was a mistake)
+or update our code if it was a sensible change. This also puts less burden on
+authors to support old, legacy versions of their software unnecessarily.
+
+Historically, we've had approximately one such breakage per year, which were all
+detected and fixed within a few hours. The cost of these small, rare,
+interruptions is much less expensive than having to periodically move every
+dependency in the project to the latest versions. Some examples of this include:
+
+* We caught the `go-bindata` swap before it was publicly known, and fixed it in:
+[adbe9c7be178898de3645b0ed17ed2ca06646017](https://github.com/purpleidea/mgmt/commit/adbe9c7be178898de3645b0ed17ed2ca06646017).
+
+* We caught the `codegangsta/cli` API change improvement, and fixed it in:
+[ab73261fd4e98cf7ecb08066ad228a8f559ba16a](https://github.com/purpleidea/mgmt/commit/ab73261fd4e98cf7ecb08066ad228a8f559ba16a).
+
+* We caught an un-announced libvirt API change, and promptly fixed it in:
+[95cb94a03958a9d2ebf01df0821a8c13a4f3a28c](https://github.com/purpleidea/mgmt/commit/95cb94a03958a9d2ebf01df0821a8c13a4f3a28c).
+
+If we choose responsible dependencies, then it usually means that those authors
+are also responsible with their changes to API and to git master. If we ever
+find that it's not the case, then we will either switch that dependency to a
+more responsible version, or fork it if necessary.
+
+Occasionally, we want to pin a dependency to a particular version. This can
+happen if the project treats `git master` as an unstable branch, or because a
+dependency needs a newer version of golang than the minimum that we require for
+our project. In those cases it's sensible to assume the technical debt, and
+vendor the dependency. The common tools such as `glide` and `godep` work by
+requiring you install their software, and by either storing a yaml file with the
+version of that dependency in your repository, and/or copying all of that code
+into git and explicitly storing it. This project thinks that all of these
+solutions are wasteful and unnecessary, particularly when an existing elegant
+solution already exists: `[git submodules](https://git-scm.com/book/en/v2/Git-Tools-Submodules)`.
+
+The advantages of using `git submodules` are three-fold:
+1. You already have the required tools installed.
+2. You only store a pointer to the dependency, not additional files or code.
+3. The git submodule tools let you easily switch dependency versions, see diff
+output, and responsibly plan and test your versions bumps with ease.
+
+Don't blindly use the tools that others tell you to. Learn what they do, think
+for yourself, and become a power user today! That process led us to using
+`git submodules`. Hopefully you'll come to the same conclusions that we did.
+
+### Did you know that there is a band named `MGMT`?
+
+I didn't realize this when naming the project, and it is accidental. After much
+anguishing, I chose the name because it was short and I thought it was
+appropriately descriptive. If you need a less ambiguous search term or phrase,
+you can try using `mgmtconfig` or `mgmt config`.
+
+It also doesn't stand for
+[Methyl Guanine Methyl Transferase](https://en.wikipedia.org/wiki/O-6-methylguanine-DNA_methyltransferase)
+which definitely existed before the band did.
+
+### You didn't answer my question, or I have a question!
+
+It's best to ask on [IRC](https://webchat.freenode.net/?channels=#mgmtconfig)
+to see if someone can help you. Once we get a big enough community going, we'll
+add a mailing list. If you don't get any response from the above, you can
+contact me through my [technical blog](https://purpleidea.com/contact/)
+and I'll do my best to help. If you have a good question, please add it as a
+patch to this documentation. I'll merge your question, and add a patch with the
+answer!

docs/function-guide.md

@@ -0,0 +1,437 @@
+# Function guide
+
+## Overview
+
+The `mgmt` tool has built-in functions which add useful, reactive functionality
+to the language. This guide describes the different function API's that are
+available. It is meant to instruct developers on how to write new functions.
+Since `mgmt` and the core functions are written in golang, some prior golang
+knowledge is assumed.
+
+## Theory
+
+Functions in `mgmt` are similar to functions in other languages, however they
+also have a [reactive](https://en.wikipedia.org/wiki/Functional_reactive_programming)
+component. Our functions can produce events over time, and there are different
+ways to write functions. For some background on this design, please read the
+[original article](https://purpleidea.com/blog/2018/02/05/mgmt-configuration-language/)
+on the subject.
+
+## Native Functions
+
+Native functions are functions which are implemented in the mgmt language
+itself. These are currently not available yet, but are coming soon. Stay tuned!
+
+## Simple Function API
+
+Most functions should be implemented using the simple function API. This API
+allows you to implement simple, static, [pure](https://en.wikipedia.org/wiki/Pure_function)
+functions that don't require you to write much boilerplate code. They will be
+automatically re-evaluated as needed when their input values change. These will
+all be automatically made available as helper functions within mgmt templates,
+and are also available for use anywhere inside mgmt programs.
+
+You'll need some basic knowledge of using the [`types`](https://github.com/purpleidea/mgmt/tree/master/lang/types)
+library which is included with mgmt. This library lets you interact with the
+available types and values in the mgmt language. It is very easy to use, and
+should be fairly intuitive. Most of what you'll need to know can be inferred
+from looking at example code.
+
+To implement a function, you'll need to create a file in
+[`lang/funcs/simple/`](https://github.com/purpleidea/mgmt/tree/master/lang/funcs/simple/).
+The function should be implemented as a `FuncValue` in our type system. It is
+then registered with the engine during `init()`. An example explains it best:
+
+### Example
+
+```golang
+package simple
+
+import (
+	"fmt"
+
+	"github.com/purpleidea/mgmt/lang/types"
+)
+
+// you must register your functions in init when the program starts up
+func init() {
+	// Example function that squares an int and prints out answer as an str.
+	Register("talkingsquare", &types.FuncValue{
+		T: types.NewType("func(a int) str"), // declare the signature
+		V: func(input []types.Value) (types.Value, error) {
+			i := input[0].Int() // get first arg as an int64
+			// must return the above specified value
+			return &types.StrValue{
+				V: fmt.Sprintf("%d^2 is %d", i, i * i),
+			}, nil // no serious errors occurred
+		},
+	})
+}
+```
+
+This simple function accepts one `int` as input, and returns one `str`.
+Functions can have zero or more inputs, and must have exactly one output. You
+must be sure to use the `types` library correctly, since if you try and access
+an input which should not exist (eg: `input[2]`, when there are only two
+that are expected), then you will cause a panic. If you have declared that a
+particular argument is an `int` but you try to read it with `.Bool()` you will
+also cause a panic. Lastly, make sure that you return a value in the correct
+type or you will also cause a panic!
+
+If anything goes wrong, you can return an error, however this will cause the
+mgmt engine to shutdown. It should be seen as the equivalent to calling a
+`panic()`, however it is safer because it brings the engine down cleanly.
+Ideally, your functions should never need to error. You should never cause a
+real `panic()`, since this could have negative consequences to the system.
+
+## Simple Polymorphic Function API
+
+Most functions should be implemented using the simple function API. If they need
+to have multiple polymorphic forms under the same name, then you can use this
+API. This is useful for situations when it would be unhelpful to name the
+functions differently, or when the number of possible signatures for the
+function would be infinite.
+
+The canonical example of this is the `len` function which returns the number of
+elements in either a `list` or a `map`. Since lists and maps are two different
+types, you can see that polymorphism is more convenient than requiring a
+`listlen` and `maplen` function. Nevertheless, it is also required because a
+`list of int` is a different type than a `list of str`, which is a different
+type than a `list of list of str` and so on. As you can see the number of
+possible input types for such a `len` function is infinite.
+
+Another downside to implementing your functions with this API is that they will
+*not* be made available for use inside templates. This is a limitation of the
+`golang` template library. In the future if this limitation proves to be
+significantly annoying, we might consider writing our own template library.
+
+As with the simple, non-polymorphic API, you can only implement [pure](https://en.wikipedia.org/wiki/Pure_function)
+functions, without writing too much boilerplate code. They will be automatically
+re-evaluated as needed when their input values change.
+
+To implement a function, you'll need to create a file in
+[`lang/funcs/simplepoly/`](https://github.com/purpleidea/mgmt/tree/master/lang/funcs/simplepoly/).
+The function should be implemented as a list of `FuncValue`'s in our type
+system. It is then registered with the engine during `init()`. You may also use
+the `variant` type in your type definitions. This special type will never be
+seen inside a running program, and will get converted to a concrete type if a
+suitable match to this signature can be found. Be warned that signatures which
+contain too many variants, or which are very general, might be hard for the
+compiler to match, and ambiguous type graphs make for user compiler errors.
+
+An example explains it best:
+
+### Example
+
+```golang
+package simplepoly
+
+import (
+	"fmt"
+
+	"github.com/purpleidea/mgmt/lang/types"
+)
+
+func init() {
+	Register("len", []*types.FuncValue{
+		{
+			T: types.NewType("func([]variant) int"),
+			V: Len,
+		},
+		{
+			T: types.NewType("func({variant: variant}) int"),
+			V: Len,
+		},
+	})
+}
+
+// Len returns the number of elements in a list or the number of key pairs in a
+// map. It can operate on either of these types.
+func Len(input []types.Value) (types.Value, error) {
+	var length int
+	switch k := input[0].Type().Kind; k {
+	case types.KindList:
+		length = len(input[0].List())
+	case types.KindMap:
+		length = len(input[0].Map())
+
+	default:
+		return nil, fmt.Errorf("unsupported kind: %+v", k)
+	}
+
+	return &types.IntValue{
+		V: int64(length),
+	}, nil
+}
+```
+
+This simple polymorphic function can accept an infinite number of signatures, of
+which there are two basic forms. Both forms return an `int` as is seen above.
+The first form takes a `[]variant` which means a `list` of `variant`'s, which
+means that it can be a list of any type, since `variant` itself is not a
+concrete type. The second form accepts a `{variant: variant}`, which means that
+it accepts any form of `map` as input.
+
+The implementation for both of these forms is the same: it is handled by the
+same `Len` function which is clever enough to be able to deal with any of the
+type signatures possible from those two patterns.
+
+At compile time, if your `mcl` code type checks correctly, a concrete type will
+be known for each and every usage of the `len` function, and specific values
+will be passed in for this code to compute the length of. As usual, make sure to
+only write safe code that will not panic! A panic is a bug. If you really cannot
+continue, then you must return an error.
+
+## Function API
+
+To implement a reactive function in `mgmt` it must satisfy the
+[`Func`](https://github.com/purpleidea/mgmt/blob/master/lang/interfaces/func.go)
+interface. Using the [Simple Function API](#simple-function-api) is preferable
+if it meets your needs. Most functions will be able to use that API. If you
+really need something more powerful, then you can use the regular function API.
+What follows are each of the method signatures and a description of each.
+
+### Default
+
+```golang
+Info() *interfaces.Info
+```
+
+This returns some information about the function. It is necessary so that the
+compiler can type check the code correctly, and know what optimizations can be
+performed. This is usually the first method which is called by the engine.
+
+#### Example
+
+```golang
+func (obj *FooFunc) Info() *interfaces.Info {
+	return &interfaces.Info{
+		Pure: true,
+		Sig:  types.NewType("func(a int) str"),
+	}
+}
+```
+
+### Init
+
+```golang
+Init(init *interfaces.Init) error
+```
+
+This is called to initialize the function. If something goes wrong, it should
+return an error. It is passed a struct that contains all the important
+information and poiinters that it might need to work with throughout its
+lifetime. As a result, it will need to save a copy to that pointer for future
+use in the other methods.
+
+#### Example
+
+```golang
+// Init runs some startup code for this function.
+func (obj *FooFunc) Init(init *interfaces.Init) error {
+	obj.init = init
+	obj.closeChan = make(chan struct{}) // shutdown signal
+	return nil
+}
+```
+
+### Close
+
+```golang
+Close() error
+```
+
+This is called to cleanup the function. It usually causes the stream to
+shutdown. Even if `Stream()` decided to shutdown early, it might still get
+called. It is usually called by the engine to tell the function to shutdown.
+
+#### Example
+
+```golang
+// Close runs some shutdown code for this function and turns off the stream.
+func (obj *FooFunc) Close() error {
+	close(obj.closeChan) // send a signal to tell the stream to close
+	return nil
+}
+```
+
+### Stream
+
+```golang
+Stream() error
+```
+
+`Stream` is where the real _work_ is done. This method is started by the
+language function engine. It will run this function while simultaneously sending
+it values on the `input` channel. It will only send a complete set of input
+values. You should send a value to the output channel when you have decided that
+one should be produced. Make sure to only use input values of the expected type
+as declared in the `Info` struct, and send values of the similarly declared
+appropriate return type. Failure to do so will may result in a panic and
+sadness.
+
+#### Example
+
+```golang
+// Stream returns the single value that was generated and then closes.
+func (obj *FooFunc) Stream() error {
+	defer close(obj.init.Output) // the sender closes
+	var result string
+	for {
+		select {
+		case input, ok := <-obj.init.Input:
+			if !ok {
+				return nil // can't output any more
+			}
+
+			ix := input.Struct()["a"].Int()
+			if ix < 0 {
+				return fmt.Errorf("we can't deal with negatives")
+			}
+
+			result = fmt.Sprintf("the input is: %d", ix)
+
+		case <-obj.closeChan:
+			return nil
+		}
+
+		select {
+		case obj.init.Output <- &types.StrValue{
+			V: result,
+		}:
+
+		case <-obj.closeChan:
+			return nil
+		}
+	}
+}
+```
+
+As you can see, we read our inputs from the `input` channel, and write to the
+`output` channel. Our code is careful to never block or deadlock, and can always
+exit if a close signal is requested. It also cleans up after itself by closing
+the `output` channel when it is done using it. This is done easily with `defer`.
+If it notices that the `input` channel closes, then it knows that no more input
+values are coming and it can consider shutting down early.
+
+## Further considerations
+
+There is some additional information that any function author will need to know.
+Each issue is listed separately below!
+
+### Function struct
+
+Each function will implement methods as pointer receivers on a function struct.
+The naming convention for resources is that they end with a `Func` suffix.
+
+#### Example
+
+```golang
+type FooFunc struct {
+	init *interfaces.Init
+
+	// this space can be used if needed
+
+	closeChan chan struct{} // shutdown signal
+}
+```
+
+### Function registration
+
+All functions must be registered with the engine so that they can be found. This
+also ensures they can be encoded and decoded. Make sure to include the following
+code snippet for this to work.
+
+```golang
+func init() { // special golang method that runs once
+	funcs.Register("foo", func() interfaces.Func { return &FooFunc{} })
+}
+```
+
+### Composite functions
+
+Composite functions are functions which import one or more existing functions.
+This is useful to prevent code duplication in higher level function scenarios.
+Unfortunately no further documentation about this subject has been written. To
+expand this section, please send a patch! Please contact us if you'd like to
+work on a function that uses this feature, or to add it to an existing one!
+We don't expect this functionality to be particularly useful or common, as it's
+probably easier and preferable to simply import common golang library code into
+multiple different functions instead.
+
+## Polymorphic Function API
+
+The polymorphic function API is an API that lets you implement functions which
+do not necessarily have a single static function signature. After compile time,
+all functions must have a static function signature. We also know that there
+might be different ways you would want to call `printf`, such as:
+`printf("the %s is %d", "answer", 42)` or `printf("3 * 2 = %d", 3 * 2)`. Since
+you couldn't implement the infinite number of possible signatures, this API lets
+you write code which can be coerced into different forms. This makes
+implementing what would appear to be generic or polymorphic, instead something
+that is actually static and that still has the static type safety properties
+that were guaranteed by the mgmt language.
+
+Since this is an advanced topic, it is not described in full at this time. For
+more information please have a look at the source code comments, some of the
+existing implementations, and ask around in the community.
+
+## Frequently asked questions
+
+(Send your questions as a patch to this FAQ! I'll review it, merge it, and
+respond by commit with the answer.)
+
+### Can I use global variables?
+
+Probably not. You must assume that multiple copies of your function may be used
+at the same time. If they require a global variable, it's likely this won't
+work. Instead it's probably better to use a struct local variable if you need to
+store some state.
+
+There might be some rare instances where a global would be acceptable, but if
+you need one of these, you're probably already an internals expert. If you think
+they need to lock or synchronize so as to not overwhelm an external resource,
+then you have to be especially careful not to cause deadlocking the mgmt engine.
+
+### Can I write functions in a different language?
+
+Currently `golang` is the only supported language for built-in functions. We
+might consider allowing external functions to be imported in the future. This
+will likely require a language that can expose a C-like API, such as `python` or
+`ruby`. Custom `golang` functions are already possible when using mgmt as a lib.
+
+### What new functions need writing?
+
+There are still many ideas for new functions that haven't been written yet. If
+you'd like to contribute one, please contact us and tell us about your idea!
+
+### Can I generate many different `FuncValue` implementations from one function?
+
+Yes, you can use a function generator in `golang` to build multiple different
+implementations from the same function generator. You just need to implement a
+function which *returns* a `golang` type of `func([]types.Value) (types.Value, error)`
+which is what `FuncValue` expects. The generator function can use any input it
+wants to build the individual functions, thus helping with code re-use.
+
+### How do I determine the signature of my simple, polymorphic function?
+
+The determination of the input portion of the function signature can be
+determined by inspecting the length of the input, and the specific type each
+value has. Length is done in the standard `golang` way, and the type of each
+element can be ascertained with the `Type()` method available on every value.
+
+Knowing the output type is trickier. If it can not be inferred in some manner,
+then the only way is to keep track of this yourself. You can use a function
+generator to build your `FuncValue` implementations, and pass in the unique
+signature to each one as you are building them. Using a generator is a common
+technique which was mentioned previously.
+
+### Where can I find more information about mgmt?
+
+Additional blog posts, videos and other material [is available!](https://github.com/purpleidea/mgmt/blob/master/docs/on-the-web.md).
+
+## Suggestions
+
+If you have any ideas for API changes or other improvements to function writing,
+please let us know! We're still pre 1.0 and pre 0.1 and happy to break API in
+order to get it right!

docs/language-guide.md

@@ -0,0 +1,639 @@
+# Language guide
+
+## Overview
+
+The `mgmt` tool has various frontends, each of which may produce a stream of
+between zero or more graphs that are passed to the engine for desired state
+application. In almost all scenarios, you're going to want to use the language
+frontend. This guide describes some of the internals of the language.
+
+## Theory
+
+The mgmt language is a declarative (immutable) functional, reactive programming
+language. It is implemented in `golang`. A longer introduction to the language
+is [available as a blog post here](https://purpleidea.com/blog/2018/02/05/mgmt-configuration-language/)!
+
+### Types
+
+All expressions must have a type. A composite type such as a list of strings
+(`[]str`) is different from a list of integers (`[]int`).
+
+There _is_ a _variant_ type in the language's type system, but it is only used
+internally and only appears briefly when needed for type unification hints
+during static polymorphic function generation. This is an advanced topic which
+is not required for normal usage of the software.
+
+The implementation of the internal types can be found in
+[lang/types/](https://github.com/purpleidea/mgmt/tree/master/lang/types/).
+
+#### bool
+
+A `true` or `false` value.
+
+#### str
+
+Any `"string!"` enclosed in quotes.
+
+#### int
+
+A number like `42` or `-13`. Integers are represented internally as golang's
+`int64`.
+
+#### float
+
+A floating point number like: `3.1415926`. Float's are represented internally as
+golang's `float64`.
+
+#### list
+
+An ordered collection of values of the same type, eg: `[6, 7, 8, 9,]`. It is
+worth mentioning that empty lists have a type, although without type hints it
+can be impossible to infer the item's type.
+
+#### map
+
+An unordered set of unique keys of the same type and corresponding value pairs
+of another type, eg:
+`{"boiling" => 100, "freezing" => 0, "room" => "25", "house" => 22, "canada" => -30,}`.
+That is to say, all of the keys must have the same type, and all of the values
+must have the same type. You can use any type for either, although it is
+probably advisable to avoid using very complex types as map keys.
+
+#### struct
+
+An ordered set of field names and corresponding values, each of their own type,
+eg: `struct{answer => "42", james => "awesome", is_mgmt_awesome => true,}`.
+These are useful for combining more than one type into the same value. Note the
+syntactical difference between these and map's: the key's in map's have types,
+and as a result, string keys are enclosed in quotes, whereas struct _fields_ are
+not string values, and as such are bare and specified without quotes.
+
+#### func
+
+An ordered set of optionally named, differently typed input arguments, and a
+return type, eg: `func(s str) int` or:
+`func(bool, []str, {str: float}) struct{foo str; bar int}`.
+
+### Expressions
+
+Expressions, and the `Expr` interface need to be better documented. For now
+please consume
+[lang/interfaces/ast.go](https://github.com/purpleidea/mgmt/tree/master/lang/interfaces/ast.go).
+These docs will be expanded on when things are more certain to be stable.
+
+### Statements
+
+There are a very small number of statements in our language. They include:
+
+- **bind**: bind's an expression to a variable within that scope
+	- eg: `$x = 42`
+- **if**: produces up to one branch of statements based on a conditional
+expression
+
+	```mcl
+	if <conditional> {
+		<statements>
+	} else {
+		# the else branch is optional for if statements
+		<statements>
+	}
+	```
+
+- **resource**: produces a resource
+
+	```mcl
+	file "/tmp/hello" {
+		content => "world",
+		mode => "o=rwx",
+	}
+	```
+
+- **edge**: produces an edge
+
+	```mcl
+	File["/tmp/hello"] -> Print["alert4"]
+	```
+
+All statements produce _output_. Output consists of between zero and more
+`edges` and `resources`. A resource statement can produce a resource, whereas an
+`if` statement produces whatever the chosen branch produces. Ultimately the goal
+of executing our programs is to produce a list of `resources`, which along with
+the produced `edges`, is built into a resource graph. This graph is then passed
+to the engine for desired state application.
+
+#### Bind
+
+This section needs better documentation.
+
+#### If
+
+This section needs better documentation.
+
+#### Resource
+
+Resources express the idempotent workloads that we want to have apply on our
+system. They correspond to vertices in a [graph](https://en.wikipedia.org/wiki/Directed_acyclic_graph)
+which represent the order in which their declared state is applied. You will
+usually want to pass in a number of parameters and associated values to the
+resource to control how it behaves. For example, setting the `content` parameter
+of a `file` resource to the string `hello`, will cause the contents of that file
+to contain the string `hello` after it has run.
+
+For some parameters, there is a distinction between an unspecified parameter,
+and a parameter with a `zero` value. For example, for the file resource, you
+might choose to set the `content` parameter to be the empty string, which would
+ensure that the file has a length of zero. Alternatively you might wish to not
+specify the file contents at all, which would leave that property undefined. If
+you omit listing a property, then it will be undefined. To control this property
+programmatically, you need to specify an `is-defined` value, as well as the
+value to use if that boolean is true. You can do this with the resource-specific
+`elvis` operator.
+
+```mcl
+$b = true # change me to false and then try editing the file manually
+file "/tmp/mgmt-elvis" {
+	content => $b ?: "hello world\n",
+	state => "exists",
+}
+```
+
+This example is static, however you can imagine that the `$b` value might be
+chosen in a programmatic way, even one in which that value varies over time. If
+it evaluates to `true`, then the parameter will be used. If no `elvis` operator
+is specified, then the parameter value will also be used. If the parameter is
+not specified, then it will obviously not be used.
+
+Resources may also declare edges internally. The edges may point to or from
+another resource, and may optionally include a notification. The four properties
+are: `Before`, `Depend`, `Notify` and `Listen`. The first two represent normal
+edge dependencies, and the second two are normal edge dependencies that also
+send notifications. You may have multiples of these per resource, including
+multiple `Depend` lines if necessary. Each of these properties also supports the
+conditional inclusion `elvis` operator as well.
+
+For example, you may write is:
+
+```mcl
+$b = true # for example purposes
+if $b {
+	pkg "drbd" {
+		state => "installed",
+
+		# multiple properties may be used in the same resource
+		Before => File["/etc/drbd.conf"],
+		Before => Svc["drbd"],
+	}
+}
+file "/etc/drbd.conf" {
+	content => "some config",
+
+	Depend => $b ?: Pkg["drbd"],
+	Notify => Svc["drbd"],
+}
+svc "drbd" {
+	state => "running",
+}
+```
+
+There are two unique properties about these edges that is different from what
+you might expect from other automation software:
+
+1. The ability to specify multiples of these properties allows you to avoid
+having to manage arrays and conditional trees of these different dependencies.
+2. The keywords all have the same length, which means your code lines up nicely.
+
+#### Edge
+
+Edges express dependencies in the graph of resources which are output. They can
+be chained as a pair, or in any greater number. For example, you may write:
+
+```mcl
+Pkg["drbd"] -> File["/etc/drbd.conf"] -> Svc["drbd"]
+```
+
+to express a relationship between three resources. The first character in the
+resource kind must be capitalized so that the parser can't ascertain
+unambiguously that we are referring to a dependency relationship.
+
+### Stages
+
+The mgmt compiler runs in a number of stages. In order of execution they are:
+* [Lexing](#lexing)
+* [Parsing](#parsing)
+* [Interpolation](#interpolation)
+* [Scope propagation](#scope-propagation)
+* [Type unification](#type-unification)
+* [Function graph generation](#function-graph-generation)
+* [Function engine creation and validation](#function-engine-creation-and-validation)
+
+All of the above needs to be done every time the source code changes. After this
+point, the [function engine runs](#function-engine-running-and-interpret) and
+produces events. On every event, we "[interpret](#function-engine-running-and-interpret)"
+which produces a resource graph. This series of resource graphs are passed
+to the engine as they are produced.
+
+What follows are some notes about each step.
+
+#### Lexing
+
+Lexing is done using [nex](https://github.com/blynn/nex). It is a pure-golang
+implementation which is similar to _Lex_ or _Flex_, but which produces golang
+code instead of C. It integrates reasonably well with golang's _yacc_ which is
+used for parsing. The token definitions are in:
+[lang/lexer.nex](https://github.com/purpleidea/mgmt/tree/master/lang/lexer.nex).
+Lexing and parsing run together by calling the `LexParse` method.
+
+#### Parsing
+
+The parser used is golang's implementation of
+[yacc](https://godoc.org/golang.org/x/tools/cmd/goyacc). The documentation is
+quite abysmal, so it's helpful to rely on the documentation from standard yacc
+and trial and error. One small advantage yacc has over standard yacc is that it
+can produce error messages from examples. The best documentation is to examine
+the source. There is a short write up available [here](https://research.swtch.com/yyerror).
+The yacc file exists at:
+[lang/parser.y](https://github.com/purpleidea/mgmt/tree/master/lang/parser.y).
+Lexing and parsing run together by calling the `LexParse` method.
+
+#### Interpolation
+
+Interpolation is used to transform the AST (which was produced from lexing and
+parsing) into one which is either identical or different. It expands strings
+which might contain expressions to be interpolated (eg: `"the answer is: ${foo}"`)
+and can be used for other scenarios in which one statement or expression would
+be better represented by a larger AST. Most nodes in the AST simply return their
+own node address, and do not modify the AST.
+
+#### Scope propagation
+
+Scope propagation passes the parent scope (starting with the top-level, built-in
+scope) down through the AST. This is necessary so that children nodes can access
+variables in the scope if needed. Most AST node's simply pass on the scope
+without making any changes. The `ExprVar` node naturally consumes scope's and
+the `StmtProg` node cleverly passes the scope through in the order expected for
+the out-of-order bind logic to work.
+
+#### Type unification
+
+Each expression must have a known type. The unpleasant option is to force the
+programmer to specify by annotation every type throughout their whole program
+so that each `Expr` node in the AST knows what to expect. Type annotation is
+allowed in situations when you want to explicitly specify a type, or when the
+compiler cannot deduce it, however, most of it can usually be inferred.
+
+For type inferrence to work, each node in the AST implements a `Unify` method
+which is able to return a list of invariants that must hold true. This starts at
+the top most AST node, and gets called through to it's children to assemble a
+giant list of invariants. The invariants can take different forms. They can
+specify that a particular expression must have a particular type, or they can
+specify that two expressions must have the same types. More complex invariants
+allow you to specify relationships between different types and expressions.
+Furthermore, invariants can allow you to specify that only one invariant out of
+a set must hold true.
+
+Once the list of invariants has been collected, they are run through an
+invariant solver. The solver can return either return successfully or with an
+error. If the solver returns successfully, it means that it has found a trivial
+mapping between every expression and it's corresponding type. At this point it
+is a simple task to run `SetType` on every expression so that the types are
+known. If the solver returns in error, it is usually due to one of two
+possibilities:
+
+1. Ambiguity
+
+	The solver does not have enough information to make a definitive or
+	unique determination about the expression to type mappings. The set of
+	invariants is ambiguous, and we cannot continue. An error will be
+	returned to the programmer. In this scenario the user will probably need
+	to add a type annotation, possibly because of a design bug in the user's
+	program.
+
+2. Conflict
+
+	The solver has conflicting information that cannot be reconciled. In
+	this situation an explicit conflict has been found. If two invariants
+	are found which both expect a particular expression to have different
+	types, then it is not possible to find a valid solution. This almost
+	always happens if the user has made a type error in their program.
+
+Only one solver currently exists, but it is possible to easily plug in an
+alternate implementation if someone more skilled in the art of solver design
+would like to propose a more logical or performant variant.
+
+#### Function graph generation
+
+At this point we have a fully type AST. The AST must now be transformed into a
+directed, acyclic graph (DAG) data structure that represents the flow of data as
+necessary for everything to be reactive. Note that this graph is *different*
+from the resource graph which is produced and sent to the engine. It is just a
+coincidence that both happen to be DAG's. (You don't freak out when you see a
+list data structure show up in more than one place, do you?)
+
+To produce this graph, each node has a `Graph` method which it can call. This
+starts at the top most node, and is called down through the AST. The edges in
+the graphs must represent the individual expression values which are passed
+from node to node. The names of the edges must match the function type argument
+names which are used in the definition of the corresponding function. These
+corresponding functions must exist for each expression node and are produced by
+calling that expression's `Func` method. These are usually called by the
+function engine during function creation and validation.
+
+#### Function engine creation and validation
+
+Finally we have a graph of the data flows. The function engine must first
+initialize which creates references to each of the necessary function
+implementations, and gets information about each one. It then needs to be type
+checked to ensure that the data flows all correctly match what is expected. If
+you were to pass an `int` to a function expecting a `bool`, this would be a
+problem. If all goes well, the program should get run shortly.
+
+#### Function engine running and interpret
+
+At this point the function engine runs. It produces a stream of events which
+cause the `Output()` method of the top-level program to run, which produces the
+list of resources and edges. These are then transformed into the resource graph
+which is passed to the engine.
+
+### Function API
+
+If you'd like to create a built-in, core function, you'll need to implement the
+function API interface named `Func`. It can be found in
+[lang/interfaces/func.go](https://github.com/purpleidea/mgmt/tree/master/lang/interfaces/func.go).
+Your function must have a specific type. For example, a simple math function
+might have a signature of `func(x int, y int) int`. As you can see, all the
+types are known _before_ compile time.
+
+A separate discussion on this matter can be found in the [function guide](function-guide.md).
+
+What follows are each of the method signatures and a description of each.
+Failure to implement the API correctly can cause the function graph engine to
+block, or the program to panic.
+
+### Info
+
+```golang
+Info() *Info
+```
+
+The Info method must return a struct containing some information about your
+function. The struct has the following type:
+
+```golang
+type Info struct {
+	Sig  *types.Type // the signature of the function, must be KindFunc
+}
+```
+
+You must implement this correctly. Other fields in the `Info` struct may be
+added in the future. This method is usually called before any other, and should
+not depend on any other method being called first. Other methods must not depend
+on this method being called first.
+
+#### Example
+
+```golang
+func (obj *FooFunc) Info() *interfaces.Info {
+	return &interfaces.Info{
+		Sig: types.NewType("func(a str, b int) float"),
+	}
+}
+```
+
+### Init
+
+```golang
+Init(*Init) error
+```
+
+Init is called by the function graph engine to create an implementation of this
+function. It is passed in a struct of the following form:
+
+```golang
+type Init struct {
+	Hostname string // uuid for the host
+	Input  chan types.Value // Engine will close `input` chan
+	Output chan types.Value // Stream must close `output` chan
+	World  resources.World
+	Debug  bool
+	Logf   func(format string, v ...interface{})
+}
+```
+
+These values and references may be used (wisely) inside your function. `Input`
+will contain a channel of input structs matching the expected input signature
+for your function. `Output` will be the channel which you must send values to
+whenever a new value should be produced. This must be done in the `Stream()`
+function. You may carefully use `World` to access functionality provided by the
+engine. You may use `Logf` to log informational messages, however there is no
+guarantee that they will be displayed to the user. `Debug` specifies whether the
+function is running in a user-requested debug mode. This might cause you to want
+to print more log messages for example. You will need to save references to any
+or all of these info fields that you wish to use in the struct implementing this
+`Func` interface. At a minimum you will need to save `Output` as a minimum of
+one value must be produced.
+
+#### Example
+
+```golang
+Please see the example functions in
+[lang/funcs/core/](https://github.com/purpleidea/mgmt/tree/master/lang/funcs/core/).
+```
+
+### Stream
+
+```golang
+Stream() error
+```
+
+Stream is called by the function engine when it is ready for your function to
+start accepting input and producing output. You must always produce at least one
+value. Failure to produce at least one value will probably cause the function
+engine to hang waiting for your output. This function must close the `Output`
+channel when it has no more values to send. The engine will close the `Input`
+channel when it has no more values to send. This may or may not influence
+whether or not you close the `Output` channel.
+
+#### Example
+
+```golang
+Please see the example functions in
+[lang/funcs/core/](https://github.com/purpleidea/mgmt/tree/master/lang/funcs/core/).
+```
+
+### Close
+
+```golang
+Close() error
+```
+
+Close asks the particular function to shutdown its `Stream()` function and
+return.
+
+#### Example
+
+```golang
+Please see the example functions in
+[lang/funcs/core/](https://github.com/purpleidea/mgmt/tree/master/lang/funcs/core/).
+```
+
+### Polymorphic Function API
+
+For some functions, it might be helpful to be able to implement a function once,
+but to have multiple polymorphic variants that can be chosen at compile time.
+For this more advanced topic, you will need to use the
+[Polymorphic Function API](#polymorphic-function-api). This will help with code
+reuse when you have a small, finite number of possible type signatures, and also
+for more complicated cases where you might have an infinite number of possible
+type signatures. (eg: `[]str`, or `[][]str`, or `[][][]str`, etc...)
+
+Suppose you want to implement a function which can assume different type
+signatures. The mgmt language does not support polymorphic types-- you must use
+static types throughout the language, however, it is legal to implement a
+function which can take different specific type signatures based on how it is
+used. For example, you might wish to add a math function which could take the
+form of `func(x int, x int) int` or `func(x float, x float) float` depending on
+the input values. You might also want to implement a function which takes an
+arbitrary number of input arguments (the number must be statically fixed at the
+compile time of your program though) and which returns a string.
+
+The `PolyFunc` interface adds additional methods which you must implement to
+satisfy such a function implementation. If you'd like to implement such a
+function, then please notify the project authors, and they will expand this
+section with a longer description of the process.
+
+#### Examples
+
+What follows are a few examples that might help you understand some of the
+language details.
+
+##### Example Foo
+
+TODO: please add an example here!
+
+##### Example Bar
+
+TODO: please add an example here!
+
+## Frequently asked questions
+
+(Send your questions as a patch to this FAQ! I'll review it, merge it, and
+respond by commit with the answer.)
+
+### What is the difference between `ExprIf` and `StmtIf`?
+
+The language contains both an `if` expression, and and `if` statement. An `if`
+expression takes a boolean conditional *and* it must contain exactly _two_
+branches (a `then` and an `else` branch) which each contain one expression. The
+`if` expression _will_ return the value of one of the two branches based on the
+conditional.
+
+#### Example:
+
+```mcl
+# this is an if expression, and both branches must exist
+$b = true
+$x = if $b {
+	42
+} else {
+	-13
+}
+```
+
+The `if` statement also takes a boolean conditional, but it may have either one
+or two branches. Branches must only directly contain statements. The `if`
+statement does not return any value, but it does produce output when it is
+evaluated. The output consists primarily of resources (vertices) and edges.
+
+#### Example:
+
+```mcl
+# this is an if statement, and in this scenario the else branch was omitted
+$b = true
+if $b {
+	file "/tmp/hello" {
+		content => "world",
+	}
+}
+```
+
+### What is the difference `types.Value.Str()` and `types.Value.String()`?
+
+In the `lang/types` library, there is a `types.Value` interface. Every value in
+our type system must implement this interface. One of the methods in this
+interface is the `String() string` method. This lets you print a representation
+of the value. You will probably never need to use this method.
+
+In addition, the `types.Value` interface implements a number of helper functions
+which return the value as an equivalent golang type. If you know that the value
+is a `bool`, you can call `x.Bool()` on it. If it's a `string` you can call
+`x.Str()`. Make sure not to call one of those type methods unless you know the
+value is of that type, or you will trigger a panic!
+
+### I created a `&ListValue{}` but it's not working!
+
+If you create a base type like `bool`, `str`, `int`, or `float`, all you need to
+do is build the `&BoolValue` and set the `V` field. Eg:
+
+```golang
+someBool := &types.BoolValue{V: true}
+```
+
+If you are building a container type like `list`, `map`, `struct`, or `func`,
+then you *also* need to specify the type of the contained values. This is
+because a list has a type of `[]str`, or `[]int`, or even `[][]foo`. Eg:
+
+```golang
+someListOfStrings := &types.ListValue{
+	T: types.NewType("[]str"),	# must match the contents!
+	V: []types.Value{
+		&types.StrValue{V: "a"},
+		&types.StrValue{V: "bb"},
+		&types.StrValue{V: "ccc"},
+	},
+}
+```
+
+If you don't build these properly, then you will cause a panic! Even empty lists
+have a type.
+
+### I don't like the mgmt language, is there an alternative?
+
+Yes, the language is just one of the available "frontends" that passes a stream
+of graphs to the engine "backend". While it _is_ the recommended way of using
+mgmt, you're welcome to either use an alternate frontend, or write your own. To
+write your own frontend, you must implement the
+[GAPI](https://github.com/purpleidea/mgmt/blob/master/gapi/gapi.go) interface.
+
+### I'm an expert in FRP, and you got it all wrong; even the names of things!
+
+I am certainly no expert in FRP, and I've certainly got lots more to learn. One
+thing FRP experts might notice is that some of the concepts from FRP are either
+named differently, or are notably absent.
+
+In mgmt, we don't talk about behaviours, events, or signals in the strict FRP
+definitons of the words. Firstly, because we only support discretized, streams
+of values with no plan to add continuous semantics. Secondly, because we prefer
+to use terms which are more natural and relatable to what our target audience is
+expecting. Our users are more likely to have a background in Physiology, or
+systems administration than a background in FRP.
+
+Having said that, we hope that the FRP community will engage with us and help
+improve the parts that we got wrong. Even if that means adding continuous
+behaviours!
+
+### This is brilliant, may I give you a high-five?
+
+Thank you, and yes, probably. "Props" may also be accepted, although patches are
+preferred. If you can't do either, [donations](https://purpleidea.com/misc/donate/)
+to support the project are welcome too!
+
+### Where can I find more information about mgmt?
+
+Additional blog posts, videos and other material
+[is available!](https://github.com/purpleidea/mgmt/blob/master/docs/on-the-web.md).
+
+## Suggestions
+
+If you have any ideas for changes or other improvements to the language, please
+let us know! We're still pre 1.0 and pre 0.1 and happy to change it in order to
+get it right!

docs/on-the-web.md

@@ -0,0 +1,45 @@
+# On the web
+
+Here is a list of places mgmt has appeared on the web. Feel free to send a patch
+if we missed something that you think is relevant!
+
+## Links
+
+| Author | Format | Subject |
+|---|---|---|
+| James Shubin | blog | [Next generation configuration mgmt](https://purpleidea.com/blog/2016/01/18/next-generation-configuration-mgmt/) |
+| James Shubin | video | [Introductory recording from DevConf.cz 2016](https://www.youtube.com/watch?v=GVhpPF0j-iE&html5=1) |
+| James Shubin | video | [Introductory recording from CfgMgmtCamp.eu 2016](https://www.youtube.com/watch?v=fNeooSiIRnA&html5=1) |
+| Julian Dunn | video | [On mgmt at CfgMgmtCamp.eu 2016](https://www.youtube.com/watch?v=kfF9IATUask&t=1949&html5=1) |
+| Walter Heck | slides | [On mgmt at CfgMgmtCamp.eu 2016](http://www.slideshare.net/olindata/configuration-management-time-for-a-4th-generation/3) |
+| Marco Marongiu | blog | [On mgmt](http://syslog.me/2016/02/15/leap-or-die/) |
+| Felix Frank | blog | [From Catalog To Mgmt (on puppet to mgmt "transpiling")](https://ffrank.github.io/features/2016/02/18/from-catalog-to-mgmt/) |
+| James Shubin | blog | [Automatic edges in mgmt (...and the pkg resource)](https://purpleidea.com/blog/2016/03/14/automatic-edges-in-mgmt/) |
+| James Shubin | blog | [Automatic grouping in mgmt](https://purpleidea.com/blog/2016/03/30/automatic-grouping-in-mgmt/) |
+| John Arundel | tweet | [“Puppet’s days are numbered.”](https://twitter.com/bitfield/status/732157519142002688) |
+| Felix Frank | blog | [Puppet, Meet Mgmt (on puppet to mgmt internals)](https://ffrank.github.io/features/2016/06/12/puppet,-meet-mgmt/) |
+| Felix Frank | blog | [Puppet Powered Mgmt (puppet to mgmt tl;dr)](https://ffrank.github.io/features/2016/06/19/puppet-powered-mgmt/) |
+| James Shubin | blog | [Automatic clustering in mgmt](https://purpleidea.com/blog/2016/06/20/automatic-clustering-in-mgmt/) |
+| James Shubin | video | [Recording from CoreOSFest 2016](https://www.youtube.com/watch?v=KVmDCUA42wc&html5=1) |
+| James Shubin | video | [Recording from DebConf16](http://meetings-archive.debian.net/pub/debian-meetings/2016/debconf16/Next_Generation_Config_Mgmt.webm) ([Slides](https://annex.debconf.org//debconf-share/debconf16/slides/15-next-generation-config-mgmt.pdf)) |
+| Felix Frank | blog | [Edging It All In (puppet and mgmt edges)](https://ffrank.github.io/features/2016/07/12/edging-it-all-in/) |
+| Felix Frank | blog | [Translating All The Things (puppet to mgmt translation warnings)](https://ffrank.github.io/features/2016/08/19/translating-all-the-things/) |
+| James Shubin | video | [Recording from systemd.conf 2016](https://www.youtube.com/watch?v=jB992Zb3nH0&html5=1) |
+| James Shubin | blog | [Remote execution in mgmt](https://purpleidea.com/blog/2016/10/07/remote-execution-in-mgmt/) |
+| James Shubin | video | [Recording from High Load Strategy 2016](https://vimeo.com/191493409) |
+| James Shubin | video | [Recording from NLUUG 2016](https://www.youtube.com/watch?v=MmpwOQAb_SE&html5=1) |
+| James Shubin | blog | [Send/Recv in mgmt](https://purpleidea.com/blog/2016/12/07/sendrecv-in-mgmt/) |
+| Julien Pivotto | blog | [Augeas resource for mgmt](https://roidelapluie.be/blog/2017/02/14/mgmt-augeas/) |
+| James Shubin | blog | [Metaparameters in mgmt](https://purpleidea.com/blog/2017/03/01/metaparameters-in-mgmt/) |
+| James Shubin | video | [Recording from Incontro DevOps 2017](https://vimeo.com/212241877) |
+| Yves Brissaud | blog | [mgmt aux HumanTalks Grenoble (french)](http://log.winsos.net/2017/04/12/mgmt-aux-human-talks-grenoble.html) |
+| James Shubin | video | [Recording from OSDC Berlin 2017](https://www.youtube.com/watch?v=LkEtBVLfygE&html5=1) |
+| Jonathan Gold | blog | [AWS:EC2 in mgmt](http://jonathangold.ca/awsec2-in-mgmt/) |
+| James Shubin | video | [Recording from OSMC Nuremberg 2017](https://www.youtube.com/watch?v=hSVadQLeplU&html5=1) |
+| James Shubin | video | [Recording from LCA 2018, Developers Miniconf](https://www.youtube.com/watch?v=OvgGfW0ilbE) |
+| James Shubin | video | [Recording from LCA 2018, Sysadmin Miniconf](https://www.youtube.com/watch?v=ELq1XOJMIPY) |
+| James Shubin | video | [Recording from LCA 2018, Main Conference](https://www.youtube.com/watch?v=_9PG64AOQ3w) |
+| James Shubin | video | [Recording from DevConf.cz 2017](https://www.youtube.com/watch?v=-FPEK08l1Zk) |
+| James Shubin | video | [Recording from FOSDEM 2018, Config Management Devroom](https://video.fosdem.org/2018/UA2.114/mgmt.webm) |
+| James Shubin | blog | [Mgmt Configuration Language](https://purpleidea.com/blog/2018/02/05/mgmt-configuration-language/) |
+| James Shubin | video | [Recording from CfgMgmtCamp.eu 2018](https://www.youtube.com/watch?v=NxObmwZDyrI) |

docs/prometheus.md

@@ -30,8 +30,9 @@ Here is a list of the metrics we provide:
 - `mgmt_resources_total`: The number of resources that mgmt is managing
 - `mgmt_checkapply_total`: The number of CheckApply's that mgmt has run
 - `mgmt_failures_total`: The number of resources that have failed
-- `mgmt_failures_current`: The number of resources that have failed
-- `mgmt_graph_start_time_seconds`: Start time of the current graph since unix epoch in seconds
+- `mgmt_failures`: The number of resources that have failed
+- `mgmt_graph_start_time_seconds`: Start time of the current graph since unix
+epoch in seconds
 
 For each metric, you will get some extra labels:
 
@@ -57,10 +58,9 @@ We do not have grafana dashboards yet. Patches welcome!
 
 - [prometheus website](https://prometheus.io/)
 - [prometheus documentation](https://prometheus.io/docs/introduction/overview/)
-- [prometheus best practices regarding metrics
-  naming](https://prometheus.io/docs/practices/naming/)
+- [prometheus best practices regarding metrics naming](https://prometheus.io/docs/practices/naming/)
 - [grafana website](http://grafana.org/)
 
 [pgc]: https://github.com/prometheus/client_golang/blob/master/prometheus/go_collector.go
 [etcdm]: https://coreos.com/etcd/docs/latest/metrics.html
-[pd]: https://github.com/prometheus/prometheus/wiki/Default-port-allocation
+[pd]: https://github.com/prometheus/prometheus/wiki/Default-port-allocations

docs/puppet-guide.md

@@ -109,8 +109,8 @@ file { "/tmp/mgmt-test":
 
 To avoid this, specify the parameter explicitly:
 
-```
-$ puppet mgmtgraph print --code 'file { "/tmp/mgmt-test": backup => false }'
+```bash
+puppet mgmtgraph print --code 'file { "/tmp/mgmt-test": backup => false }'
 ```
 
 This is tedious in a more complex manifest. A good simplification is the

docs/quick-start-guide.md

@@ -1,94 +1,182 @@
 # Quick start guide
 
 ## Introduction
+
 This guide is intended for developers. Once `mgmt` is minimally viable, we'll
-publish a quick start guide for users too. In the meantime, please contribute!
-If you're brand new to `mgmt`, it's probably a good idea to start by reading the
-[introductory article](https://ttboj.wordpress.com/2016/01/18/next-generation-configuration-mgmt/)
-or to watch an [introductory video](https://github.com/purpleidea/mgmt/#on-the-web).
+publish a quick start guide for users too. If you're brand new to `mgmt`, it's
+probably a good idea to start by reading the
+[introductory article](https://purpleidea.com/blog/2016/01/18/next-generation-configuration-mgmt/)
+or to watch an [introductory video](https://www.youtube.com/watch?v=LkEtBVLfygE&html5=1).
 Once you're familiar with the general idea, please start hacking...
 
-## Vagrant
-If you would like to avoid doing the following steps manually, we have prepared
-a [Vagrant](https://www.vagrantup.com/) environment for your convenience. From
-the project directory, run a `vagrant up`, and then a `vagrant status`. From
-there, you can `vagrant ssh` into the `mgmt` machine. The MOTD will explain the
-rest.
-
-## Dependencies
-Software projects have a few different kinds of dependencies. There are _build_
-dependencies, _runtime_ dependencies, and additionally, a few extra dependencies
-required for running the _test_ suite.
-
-### Build
-* `golang` 1.8 or higher (required, available in some distros and distributed
-  as a binary officially by [golang.org](https://golang.org/dl/))
-* golang libraries (required, available with `go get ./...`) a partial list includes:
-```
-github.com/coreos/etcd/client
-gopkg.in/yaml.v2
-gopkg.in/fsnotify.v1
-github.com/urfave/cli
-github.com/coreos/go-systemd/dbus
-github.com/coreos/go-systemd/util
-github.com/libvirt/libvirt-go
-```
-* `stringer` (optional), available as a package on some platforms, otherwise via `go get`
-```
-golang.org/x/tools/cmd/stringer
-```
-* `pandoc` (optional), for building a pdf of the documentation
-
-### Runtime
-A relatively modern GNU/Linux system should be able to run `mgmt` without any
-problems. Since `mgmt` runs as a single statically compiled binary, all of the
-library dependencies are included. It is expected, that certain advanced
-resources require host specific facilities to work. These requirements are
-listed below:
-
-| Resource | Dependency        | Version |
-|----------|-------------------|---------|
-| file     | inotify           | ?       |
-| hostname | systemd-hostnamed | ?       |
-| nspawn   | systemd-nspawn    | ?       |
-| pkg      | packagekitd       | ?       |
-| svc      | systemd           | ?       |
-| virt     | libvirtd          | ?       |
-
-For building a visual representation of the graph, `graphviz` is required.
-
-### Testing
-* golint `github.com/golang/lint/golint`
-
 ## Quick start
-* Make sure you have golang version 1.8 or greater installed.
+
+### Installing golang
+
+* You need golang version 1.9 or greater installed.
+	* To install on rpm style systems: `sudo dnf install golang`
+	* To install on apt style systems: `sudo apt install golang`
+	* To install on macOS systems install [Homebrew](https://brew.sh)
+	and run: `brew install go`
+* You can run `go version` to check the golang version.
+* If your distro is tool old, you may need to [download](https://golang.org/dl/)
+a newer golang version.
+
+### Setting up golang
+
 * If you do not have a GOPATH yet, create one and export it:
+
 ```
 mkdir $HOME/gopath
 export GOPATH=$HOME/gopath
 ```
+
 * You might also want to add the GOPATH to your `~/.bashrc` or `~/.profile`.
 * For more information you can read the [GOPATH documentation](https://golang.org/cmd/go/#hdr-GOPATH_environment_variable).
-* Next download the mgmt code base, and switch to that directory:
+
+### Getting the mgmt code and dependencies
+
+* Download the `mgmt` code into the GOPATH, and switch to that directory:
+
 ```
 mkdir -p $GOPATH/src/github.com/purpleidea/
 cd $GOPATH/src/github.com/purpleidea/
 git clone --recursive https://github.com/purpleidea/mgmt/
 cd $GOPATH/src/github.com/purpleidea/mgmt
 ```
-* Run `make deps` to install system and golang dependencies. Take a look at `misc/make-deps.sh` for details.
+
+* Add $GOPATH/bin to $PATH
+
+```
+export PATH=$PATH:$GOPATH/bin
+```
+
+* Run `make deps` to install system and golang dependencies. Take a look at
+`misc/make-deps.sh` for details.
 * Run `make build` to get a freshly built `mgmt` binary.
-* Run `time ./mgmt run --yaml examples/graph0.yaml --converged-timeout=5 --tmp-prefix` to try out a very simple example!
-* To run continuously in the default mode of operation, omit the `--converged-timeout` option.
-* Have fun hacking on our future technology!
+
+### Running mgmt
+
+* Run `time ./mgmt run --lang examples/lang/hello0.mcl --tmp-prefix` to try out
+a very simple example!
+* Look in that example file that you ran to see if you can figure out what it
+did!
+* Have fun hacking on our future technology and get involved to shape the
+project!
 
 ## Examples
-Please look in the [examples/](../examples/) folder for some examples!
 
-## Installation
+Please look in the [examples/lang/](../examples/lang/) folder for some more
+examples!
+
+## Vagrant
+
+If you would like to avoid doing the above steps manually, we have prepared a
+[Vagrant](https://www.vagrantup.com/) environment for your convenience. From the
+project directory, run a `vagrant up`, and then a `vagrant status`. From there,
+you can `vagrant ssh` into the `mgmt` machine. The MOTD will explain the rest.
+
+## Using Docker
+
+Alternatively, you can check out the [docker-guide](docs/docker-guide.md) in
+order to develop or deploy using docker.
+
+## Information about dependencies
+
+Software projects have a few different kinds of dependencies. There are _build_
+dependencies, _runtime_ dependencies, and additionally, a few extra dependencies
+required for running the _test_ suite.
+
+### Build
+
+* `golang` 1.9 or higher (required, available in some distros and distributed
+as a binary officially by [golang.org](https://golang.org/dl/))
+
+### Runtime
+
+A relatively modern GNU/Linux system should be able to run `mgmt` without any
+problems. Since `mgmt` runs as a single statically compiled binary, all of the
+library dependencies are included. It is expected, that certain advanced
+resources require host specific facilities to work. These requirements are
+listed below:
+
+| Resource | Dependency        | Version                     | Check version with                                        |
+|----------|-------------------|-----------------------------|-----------------------------------------------------------|
+| augeas   | augeas-devel      | `augeas 1.6` or greater     | `dnf info augeas-devel` or `apt-cache show libaugeas-dev` |
+| file     | inotify           | `Linux 2.6.27` or greater   | `uname -a`                                                |
+| hostname | systemd-hostnamed | `systemd 25` or greater     | `systemctl --version`                                     |
+| nspawn   | systemd-nspawn    | `systemd ???` or greater    | `systemctl --version`                                     |
+| pkg      | packagekitd       | `packagekit 1.x` or greater | `pkcon --version`                                         |
+| svc      | systemd           | `systemd ???` or greater    | `systemctl --version`                                     |
+| virt     | libvirt-devel     | `libvirt 1.2.0` or greater  | `dnf info libvirt-devel` or `apt-cache show libvirt-dev`  |
+| virt     | libvirtd          | `libvirt 1.2.0` or greater  | `libvirtd --version`                                      |
+
+For building a visual representation of the graph, `graphviz` is required.
+
+To build `mgmt` without augeas support please run:
+`GOTAGS='noaugeas' make build`
+
+To build `mgmt` without libvirt support please run:
+`GOTAGS='novirt' make build`
+
+To build `mgmt` without augeas or libvirt support please run:
+`GOTAGS='noaugeas novirt' make build`
+
+## Binary Package Installation
+
 Installation of `mgmt` from distribution packages currently needs improvement.
+They are not always up-to-date with git master and as such are not recommended.
 At the moment we have:
 * [COPR](https://copr.fedoraproject.org/coprs/purpleidea/mgmt/)
 * [Arch](https://aur.archlinux.org/packages/mgmt/)
 
 Please contribute more! We'd especially like to see a Debian package!
+
+## OSX/macOS/Darwin development
+
+Developing and running `mgmt` on macOS is currently not supported (but not
+discouraged either). Meaning it might work but in the case it doesn't you would
+have to provide your own patches to fix problems (the project maintainer and
+community are glad to assist where needed).
+
+There are currently some issues that make `mgmt` less suitable to run for provisioning
+macOS. But as a client to provision remote servers it should run fine.
+
+Since the primary supported systems are Linux and these are the environments
+tested for it is wise to run these suites during macOS development as well. To
+ease this Docker can be leveraged ([Docker for Mac](https://docs.docker.com/docker-for-mac/)).
+
+Before running any of the commands below create the development Docker image:
+
+```
+docker/scripts/build-development
+```
+
+This image requires updating every time dependencies (`make-deps.sh`) change.
+
+Then to run the test suite:
+
+```
+docker run --rm -ti \
+	-v $PWD:/go/src/github.com/purpleidea/mgmt/ \
+	-w /go/src/github.com/purpleidea/mgmt/ \
+	purpleidea/mgmt:development \
+	make test
+```
+
+For convenience this command is wrapped in `docker/scripts/exec-development`.
+
+Basically any command can be executed this way. Because the repository source is
+mounted into the Docker container invocation will be quick and allow rapid
+testing, example:
+
+```
+docker/scripts/exec-development test/test-shell.sh load0.sh
+```
+
+Other examples:
+
+```
+docker/scripts/exec-development make build
+docker/scripts/exec-development ./mgmt run --tmp-prefix --lang examples/lang/load0.mcl
+```

docs/resource-guide.md

@@ -16,7 +16,7 @@ Resources in `mgmt` are similar to resources in other systems in that they are
 uniquely different in that they can detect when their state has changed, and as
 a result can run to revert or repair this change instantly. For some background
 on this design, please read the
-[original article](https://ttboj.wordpress.com/2016/01/18/next-generation-configuration-mgmt/)
+[original article](https://purpleidea.com/blog/2016/01/18/next-generation-configuration-mgmt/)
 on the subject.
 
 ## Resource API
@@ -27,6 +27,7 @@ interface. What follows are each of the method signatures and a description of
 each.
 
 ### Default
+
 ```golang
 Default() Res
 ```
@@ -36,6 +37,7 @@ values which already have the correct default as the golang zero value. In
 general it is preferable if the zero values make for the correct defaults.
 
 #### Example
+
 ```golang
 // Default returns some sensible defaults for this resource.
 func (obj *FooRes) Default() Res {
@@ -46,6 +48,7 @@ func (obj *FooRes) Default() Res {
 ```
 
 ### Validate
+
 ```golang
 Validate() error
 ```
@@ -57,6 +60,7 @@ quite large, it might be an indication that you should reconsider the parameter
 list and interface to this resource. This method is called _before_ `Init`.
 
 #### Example
+
 ```golang
 // Validate reports any problems with the struct definition.
 func (obj *FooRes) Validate() error {
@@ -68,6 +72,7 @@ func (obj *FooRes) Validate() error {
 ```
 
 ### Init
+
 ```golang
 Init() error
 ```
@@ -77,6 +82,7 @@ return an error. It should do any resource specific work, and finish by calling
 the `Init` method of the base resource.
 
 #### Example
+
 ```golang
 // Init initializes the Foo resource.
 func (obj *FooRes) Init() error {
@@ -95,6 +101,7 @@ shouldn't allow `Init` to dangerously `rm -rf /$the_world` if your code only
 checks `$the_world` in `Validate`. Remember to always program safely!
 
 ### Close
+
 ```golang
 Close() error
 ```
@@ -104,6 +111,7 @@ can be useful if you'd like to properly close a persistent connection that you
 opened in the `Init` method and were using throughout the resource.
 
 #### Example
+
 ```golang
 // Close runs some cleanup code for this resource.
 func (obj *FooRes) Close() error {
@@ -125,6 +133,7 @@ method! If you plan to return early if you hit an internal error, then at least
 call it with a defer!
 
 ### CheckApply
+
 ```golang
 CheckApply(apply bool) (checkOK bool, err error)
 ```
@@ -150,6 +159,7 @@ facility will detect the change, ultimately resulting in a subsequent call to
 `CheckApply`.
 
 #### Example
+
 ```golang
 // CheckApply does the idempotent work of checking and applying resource state.
 func (obj *FooRes) CheckApply(apply bool) (bool, error) {
@@ -171,6 +181,7 @@ the documentation in case you are confused as to why a debug message you've
 added to the code isn't always printed.
 
 #### Refresh notifications
+
 Some resources may choose to support receiving refresh notifications. In general
 these should be avoided if possible, but nevertheless, they do make sense in
 certain situations. Resources that support these need to verify if one was sent
@@ -184,6 +195,7 @@ have enabled their propagation. Resources that currently perform some refresh
 action include `svc`, `timer`, and `password`.
 
 #### Paired execution
+
 For many resources it is not uncommon to see `CheckApply` run twice in rapid
 succession. This is usually not a pathological occurrence, but rather a healthy
 pattern which is a consequence of the event system. When the state of the
@@ -193,6 +205,7 @@ trigger the `Watch` code! In response, a second `CheckApply` is triggered, which
 will likely find the state to now be correct.
 
 #### Summary
+
 * Anytime an error occurs during `CheckApply`, you should return `(false, err)`.
 * If the state is correct and no changes are needed, return `(true, nil)`.
 * You should only make changes to the system if `apply` is set to `true`.
@@ -200,6 +213,7 @@ will likely find the state to now be correct.
 * Returning `(true, err)` is a programming error and will cause a `Fatal`.
 
 ### Watch
+
 ```golang
 Watch() error
 ```
@@ -229,6 +243,7 @@ executed. As a result, the resource must still work even if the main loop is not
 running.
 
 #### Select
+
 The lifetime of most resources `Watch` method should be spent in an infinite
 loop that is bounded by a `select` call. The `select` call is the point where
 our method hands back control to the engine (and the kernel) so that we can
@@ -237,6 +252,7 @@ events from the engine via the `<-obj.Events()` call, and receive events for our
 resource itself!
 
 #### Events
+
 If we receive an internal event from the `<-obj.Events()` method, we can read it
 with the ReadEvent helper function. This function tells us if we should shutdown
 our resource, and if we should generate an event. When we want to send an event,
@@ -245,6 +261,7 @@ state as `dirty` if we believe it might have changed. We do this with the
 `StateOK(false)` function.
 
 #### Startup
+
 Once the `Watch` function has finished starting up successfully, it is important
 to generate one event to notify the `mgmt` engine that we're now listening
 successfully, so that it can run an initial `CheckApply` to ensure we're safely
@@ -252,6 +269,7 @@ tracking a healthy state and that we didn't miss anything when `Watch` was down
 or from before `mgmt` was running. It does this by calling the `Running` method.
 
 #### Converged
+
 The engine might be asked to shutdown when the entire state of the system has
 not seen any changes for some duration of time. The engine can determine this
 automatically, but each resource can block this if it is absolutely necessary.
@@ -270,6 +288,7 @@ prove to be useful if a resource wants to start off a long operation, but avoid
 sending out erroneous `Event` messages to keep things alive until it finishes.
 
 #### Example
+
 ```golang
 // Watch is the listener and main loop for this resource.
 func (obj *FooRes) Watch() error {
@@ -317,6 +336,7 @@ func (obj *FooRes) Watch() error {
 ```
 
 #### Summary
+
 * Remember to call the appropriate `converger` methods throughout the resource.
 * Remember to call `Startup` when the `Watch` is running successfully.
 * Remember to process internal events and shutdown promptly if asked to.
@@ -324,6 +344,7 @@ func (obj *FooRes) Watch() error {
 * Have a look at the existing resources for a rough idea of how this all works.
 
 ### Compare
+
 ```golang
 Compare(Res) bool
 ```
@@ -341,6 +362,7 @@ particular if they store some generated state, or if they aren't significant in
 some way.
 
 #### Example
+
 ```golang
 // Compare two resources and return if they are equivalent.
 func (obj *FooRes) Compare(r Res) bool {
@@ -368,6 +390,7 @@ func (obj *FooRes) Compare(r Res) bool {
 ```
 
 ### UIDs
+
 ```golang
 UIDs() []ResUID
 ```
@@ -377,6 +400,7 @@ particular resource uniquely. This is used with the AutoEdges API to determine
 if another resource can match a dependency to this one.
 
 ### AutoEdges
+
 ```golang
 AutoEdges() (AutoEdge, error)
 ```
@@ -386,6 +410,7 @@ is used to match other resources that might be relevant dependencies for this
 resource.
 
 ### CollectPattern
+
 ```golang
 CollectPattern() string
 ```
@@ -393,6 +418,7 @@ CollectPattern() string
 This is currently a stub and will be updated once the DSL is further along.
 
 ### UnmarshalYAML
+
 ```golang
 UnmarshalYAML(unmarshal func(interface{}) error) error // optional
 ```
@@ -406,6 +432,7 @@ The signature intentionally matches what is required to satisfy the `go-yaml`
 [Unmarshaler](https://godoc.org/gopkg.in/yaml.v2#Unmarshaler) interface.
 
 #### Example
+
 ```golang
 // UnmarshalYAML is the custom unmarshal handler for this struct.
 // It is primarily useful for setting the defaults.
@@ -429,10 +456,12 @@ func (obj *FooRes) UnmarshalYAML(unmarshal func(interface{}) error) error {
 ```
 
 ## Further considerations
+
 There is some additional information that any resource writer will need to know.
 Each issue is listed separately below!
 
 ### Resource struct
+
 Each resource will implement methods as pointer receivers on a resource struct.
 The resource struct must include an anonymous reference to the `BaseRes` struct.
 The naming convention for resources is that they end with a `Res` suffix. If
@@ -440,6 +469,7 @@ you'd like your resource to be accessible by the `YAML` graph API (GAPI), then
 you'll need to include the appropriate YAML fields as shown below.
 
 #### Example
+
 ```golang
 type FooRes struct {
 	BaseRes `yaml:",inline"` // base properties
@@ -453,6 +483,7 @@ type FooRes struct {
 ```
 
 ### Resource registration
+
 All resources must be registered with the engine so that they can be found. This
 also ensures they can be encoded and decoded. Make sure to include the following
 code snippet for this to work.
@@ -465,23 +496,25 @@ func init() { // special golang method that runs once
 ```
 
 ## Automatic edges
-Automatic edges in `mgmt` are well described in [this article](https://ttboj.wordpress.com/2016/03/14/automatic-edges-in-mgmt/).
+
+Automatic edges in `mgmt` are well described in [this article](https://purpleidea.com/blog/2016/03/14/automatic-edges-in-mgmt/).
 The best example of this technique can be seen in the `svc` resource.
 Unfortunately no further documentation about this subject has been written. To
 expand this section, please send a patch! Please contact us if you'd like to
 work on a resource that uses this feature, or to add it to an existing one!
 
 ## Automatic grouping
-Automatic grouping in `mgmt` is well described in [this article](https://ttboj.wordpress.com/2016/03/30/automatic-grouping-in-mgmt/).
+
+Automatic grouping in `mgmt` is well described in [this article](https://purpleidea.com/blog/2016/03/30/automatic-grouping-in-mgmt/).
 The best example of this technique can be seen in the `pkg` resource.
 Unfortunately no further documentation about this subject has been written. To
 expand this section, please send a patch! Please contact us if you'd like to
 work on a resource that uses this feature, or to add it to an existing one!
 
-
 ## Send/Recv
+
 In `mgmt` there is a novel concept called _Send/Recv_. For some background,
-please [read the introductory article](https://ttboj.wordpress.com/2016/12/07/sendrecv-in-mgmt/).
+please [read the introductory article](https://purpleidea.com/blog/2016/12/07/sendrecv-in-mgmt/).
 When using this feature, the engine will automatically send the user specified
 value to the intended destination without requiring any resource specific code.
 Any time that one of the destination values is changed, the engine automatically
@@ -523,6 +556,7 @@ such as for cache invalidation.
 Remember, `Send/Recv` only changes your resource code if you cache state.
 
 ## Composite resources
+
 Composite resources are resources which embed one or more existing resources.
 This is useful to prevent code duplication in higher level resource scenarios.
 The best example of this technique can be seen in the `nspawn` resource which
@@ -532,24 +566,70 @@ expand this section, please send a patch! Please contact us if you'd like to
 work on a resource that uses this feature, or to add it to an existing one!
 
 ## Frequently asked questions
+
 (Send your questions as a patch to this FAQ! I'll review it, merge it, and
 respond by commit with the answer.)
 
 ### Can I write resources in a different language?
+
 Currently `golang` is the only supported language for built-in resources. We
 might consider allowing external resources to be imported in the future. This
 will likely require a language that can expose a C-like API, such as `python` or
 `ruby`. Custom `golang` resources are already possible when using mgmt as a lib.
 Higher level resource collections will be possible once the `mgmt` DSL is ready.
 
+### Why does the resource API have `CheckApply` instead of two separate methods?
+
+In an early version we actually had both "parts" as separate methods, namely:
+`StateOK` (Check) and `Apply`, but the [decision](58f41eddd9c06b183f889f15d7c97af81b0331cc)
+was made to merge the two into a single method. There are two reasons for this:
+
+1. Many situations would involve the engine running both `Check` and `Apply`. If
+the resource needed to share some state (for efficiency purposes) between the
+two calls, this is much more difficult. A common example is that a resource
+might want to open a connection to `dbus` or `http` to do resource state testing
+and applying. If the methods are combined, there's no need to open and close
+them twice. A counter argument might be that you could open the connection in
+`Init`, and close it in `Close`, however you might not want that open for the
+full lifetime of the resource if you only change state occasionally.
+2. Suppose you came up with a really good reason why you wanted the two methods
+to be separate. It turns out that the current `CheckApply` can wrap this easily.
+It would look approximately like this:
+
+```golang
+func (obj *FooRes) CheckApply(apply bool) (bool, error) {
+	// my private split implementation of check and apply
+	if c, err := obj.check(); err != nil {
+		return false, err // we errored
+	} else if c {
+		return true, nil // state was good!
+	}
+
+	if !apply {
+		return false, nil // state needs fixing, but apply is false
+	}
+
+	err := obj.apply() // errors if failure or unable to apply
+
+	return false, err // always return false, with an optional error
+}
+```
+
+Feel free to use this pattern if you're convinced it's necessary. Alternatively,
+if you think I got the `Res` API wrong and you have an improvement, please let
+us know!
+
 ### What new resource primitives need writing?
+
 There are still many ideas for new resources that haven't been written yet. If
 you'd like to contribute one, please contact us and tell us about your idea!
 
 ### Where can I find more information about mgmt?
-Additional blog posts, videos and other material [is available!](https://github.com/purpleidea/mgmt/#on-the-web).
+
+Additional blog posts, videos and other material [is available!](https://github.com/purpleidea/mgmt/blob/master/docs/on-the-web.md).
 
 ## Suggestions
+
 If you have any ideas for API changes or other improvements to resource writing,
 please let us know! We're still pre 1.0 and pre 0.1 and happy to break API in
 order to get it right!

docs/resources.md

@@ -0,0 +1,177 @@
+# Resources
+
+Here we list all the built-in resources and their properties. The resource
+primitives in `mgmt` are typically more powerful than resources in other
+configuration management systems because they can be event based which lets them
+respond in real-time to converge to the desired state. This property allows you
+to build more complex resources that you probably hadn't considered in the past.
+
+In addition to the resource specific properties, there are resource properties
+(otherwise known as parameters) which can apply to every resource. These are
+called [meta parameters](documentation.md#meta-parameters) and are listed
+separately. Certain meta parameters aren't very useful when combined with
+certain resources, but in general, it should be fairly obvious, such as when
+combining the `noop` meta parameter with the [Noop](#Noop) resource.
+
+You might want to look at the [generated documentation](https://godoc.org/github.com/purpleidea/mgmt/resources)
+for more up-to-date information about these resources.
+
+* [Augeas](#Augeas): Manipulate files using augeas.
+* [Exec](#Exec): Execute shell commands on the system.
+* [File](#File): Manage files and directories.
+* [Hostname](#Hostname): Manages the hostname on the system.
+* [KV](#KV): Set a key value pair in our shared world database.
+* [Msg](#Msg): Send log messages.
+* [Noop](#Noop): A simple resource that does nothing.
+* [Nspawn](#Nspawn): Manage systemd-machined nspawn containers.
+* [Password](#Password): Create random password strings.
+* [Pkg](#Pkg):  Manage system packages with PackageKit.
+* [Svc](#Svc): Manage system systemd services.
+* [Timer](#Timer): Manage system systemd services.
+* [Virt](#Virt): Manage virtual machines with libvirt.
+
+## Augeas
+
+The augeas resource uses [augeas](http://augeas.net/) commands to manipulate
+files.
+
+## Exec
+
+The exec resource can execute commands on your system.
+
+## File
+
+The file resource manages files and directories. In `mgmt`, directories are
+identified by a trailing slash in their path name. File have no such slash.
+
+It has the following properties:
+
+* `path`: file path (directories have a trailing slash here)
+* `content`: raw file content
+* `state`: either `exists` (the default value) or `absent`
+* `mode`: octal unix file permissions
+* `owner`: username or uid for the file owner
+* `group`: group name or gid for the file group
+
+### Path
+
+The path property specifies the file or directory that we are managing.
+
+### Content
+
+The content property is a string that specifies the desired file contents.
+
+### Source
+
+The source property points to a source file or directory path that we wish to
+copy over and use as the desired contents for our resource.
+
+### State
+
+The state property describes the action we'd like to apply for the resource. The
+possible values are: `exists` and `absent`.
+
+### Recurse
+
+The recurse property limits whether file resource operations should recurse into
+and monitor directory contents with a depth greater than one.
+
+### Force
+
+The force property is required if we want the file resource to be able to change
+a file into a directory or vice-versa. If such a change is needed, but the force
+property is not set to `true`, then this file resource will error.
+
+## Hostname
+
+The hostname resource manages static, transient/dynamic and pretty hostnames
+on the system and watches them for changes.
+
+### static_hostname
+
+The static hostname is the one configured in /etc/hostname or a similar
+file.
+It is chosen by the local user. It is not always in sync with the current
+host name as returned by the gethostname() system call.
+
+### transient_hostname
+
+The transient / dynamic hostname is the one configured via the kernel's
+sethostbyname().
+It can be different from the static hostname in case DHCP or mDNS have been
+configured to change the name based on network information.
+
+### pretty_hostname
+
+The pretty hostname is a free-form UTF8 host name for presentation to the user.
+
+### hostname
+
+Hostname is the fallback value for all 3 fields above, if only `hostname` is
+specified, it will set all 3 fields to this value.
+
+## KV
+
+The KV resource sets a key and value pair in the global world database. This is
+quite useful for setting a flag after a number of resources have run. It will
+ignore database updates to the value that are greater in compare order than the
+requested key if the `SkipLessThan` parameter is set to true. If we receive a
+refresh, then the stored value will be reset to the requested value even if the
+stored value is greater.
+
+### Key
+
+The string key used to store the key.
+
+### Value
+
+The string value to set. This can also be set via Send/Recv.
+
+### SkipLessThan
+
+If this parameter is set to `true`, then it will ignore updating the value as
+long as the database versions are greater than the requested value. The compare
+operation used is based on the `SkipCmpStyle` parameter.
+
+### SkipCmpStyle
+
+By default this converts the string values to integers and compares them as you
+would expect.
+
+## Msg
+
+The msg resource sends messages to the main log, or an external service such
+as systemd's journal.
+
+## Noop
+
+The noop resource does absolutely nothing. It does have some utility in testing
+`mgmt` and also as a placeholder in the resource graph.
+
+## Nspawn
+
+The nspawn resource is used to manage systemd-machined style containers.
+
+## Password
+
+The password resource can generate a random string to be used as a password. It
+will re-generate the password if it receives a refresh notification.
+
+## Pkg
+
+The pkg resource is used to manage system packages. This resource works on many
+different distributions because it uses the underlying packagekit facility which
+supports different backends for different environments. This ensures that we
+have great Debian (deb/dpkg) and Fedora (rpm/dnf) support simultaneously.
+
+## Svc
+
+The service resource is still very WIP. Please help us my improving it!
+
+## Timer
+
+This resource needs better documentation. Please help us my improving it!
+
+## Virt
+
+The virt resource can manage virtual machines via libvirt.

docs/style-guide.md

@@ -0,0 +1,96 @@
+# Style guide
+
+## Overview
+
+This document aims to be a reference for the desired style for patches to mgmt.
+In particular it describes conventions which we use which are not officially
+enforced by the `gofmt` tool, and which might not be clearly defined elsewhere.
+Most of these are common sense to seasoned programmers, and we hope this will be
+a useful reference for new programmers.
+
+There are a lot of useful code review comments described
+[here](https://github.com/golang/go/wiki/CodeReviewComments). We don't
+necessarily follow everything strictly, but it is in general a very good guide.
+
+## Basics
+
+* All of our golang code is formatted with `gofmt`.
+
+## Comments
+
+All of our code is commented with the minimums required for `godoc` to function,
+and so that our comments pass `golint`. Code comments should either be full
+sentences (which end with a period, use proper punctuation, and capitalize the
+first word when it is not a lower cased identifier), or are short one-line
+comments in the source which are not full sentences and don't end with a period.
+
+They should explain algorithms, describe non-obvious behaviour, or situations
+which would otherwise need explanation or additional research during a code
+review. Notes about use of unfamiliar API's is a good idea for a code comment.
+
+### Example
+
+Here you can see a function with the correct `godoc` string. The first word must
+match the name of the function. It is _not_ capitalized because the function is
+private.
+
+```golang
+// square multiplies the input integer by itself and returns this product.
+func square(x int) int {
+	return x * x // we don't care about overflow errors
+}
+```
+
+## Line length
+
+In general we try to stick to 80 character lines when it is appropriate. It is
+almost *always* appropriate for function `godoc` comments and most longer
+paragraphs. Exceptions are always allowed based on the will of the maintainer.
+
+It is usually better to exceed 80 characters than to break code unnecessarily.
+If your code often exceeds 80 characters, it might be an indication that it
+needs refactoring.
+
+Occasionally inline, two line source code comments are used within a function.
+These should usually be balanced so that you don't have one line with 78
+characters and the second with only four. Split the comment between the two.
+
+## Method receiver naming
+
+[Contrary](https://github.com/golang/go/wiki/CodeReviewComments#receiver-names)
+to the specialized naming of the method receiver variable, we usually name all
+of these `obj` for ease of code copying throughout the project, and for faster
+identification when reviewing code. Some anecdotal studies have shown that it
+makes the code easier to read since you don't need to remember the name of the
+method receiver variable in each different method. This is very similar to what
+is done in `python`.
+
+### Example
+
+```golang
+// Bar does a thing, and returns the number of baz results found in our
+database.
+func (obj *Foo) Bar(baz string) int {
+	if len(obj.s) > 0 {
+		return strings.Count(obj.s, baz)
+	}
+	return -1
+}
+```
+
+## Consistent ordering
+
+In general we try to preserve a logical ordering in source files which usually
+matches the common order of execution that a _lazy evaluator_ would follow.
+
+This is also the order which is recommended when creating interface types. When
+implementing an interface, arrange your methods in the same order that they are
+declared in the interface.
+
+When implementing code for the various types in the language, please follow this
+order: `bool`, `str`, `int`, `float`, `list`, `map`, `struct`, `func`.
+
+## Suggestions
+
+If you have any ideas for suggestions or other improvements to this guide,
+please let us know!

etcd/client.go

@@ -0,0 +1,94 @@
+// Mgmt
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
+// Written by James Shubin <james@shubin.ca> and the project contributors
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+package etcd
+
+import (
+	"time"
+
+	etcd "github.com/coreos/etcd/clientv3" // "clientv3"
+	errwrap "github.com/pkg/errors"
+	context "golang.org/x/net/context"
+)
+
+// ClientEtcd provides a simple etcd client for deploy and status operations.
+type ClientEtcd struct {
+	Seeds []string // list of endpoints to try to connect
+
+	client *etcd.Client
+}
+
+// GetClient returns a handle to the raw etcd client object.
+func (obj *ClientEtcd) GetClient() *etcd.Client {
+	return obj.client
+}
+
+// GetConfig returns the config struct to be used for the etcd client connect.
+func (obj *ClientEtcd) GetConfig() etcd.Config {
+	cfg := etcd.Config{
+		Endpoints: obj.Seeds,
+		// RetryDialer chooses the next endpoint to use
+		// it comes with a default dialer if unspecified
+		DialTimeout: 5 * time.Second,
+	}
+	return cfg
+}
+
+// Connect connects the client to a server, and then builds the *API structs.
+// If reconnect is true, it will force a reconnect with new config endpoints.
+func (obj *ClientEtcd) Connect() error {
+	if obj.client != nil { // memoize
+		return nil
+	}
+
+	var err error
+	cfg := obj.GetConfig()
+	obj.client, err = etcd.New(cfg) // connect!
+	if err != nil {
+		return errwrap.Wrapf(err, "client connect error")
+	}
+	return nil
+}
+
+// Destroy cleans up the entire etcd client connection.
+func (obj *ClientEtcd) Destroy() error {
+	err := obj.client.Close()
+	//obj.wg.Wait()
+	return err
+}
+
+// Get runs a get on the client connection. This has the same signature as our
+// EmbdEtcd Get function.
+func (obj *ClientEtcd) Get(path string, opts ...etcd.OpOption) (map[string]string, error) {
+	resp, err := obj.client.Get(context.TODO(), path, opts...)
+	if err != nil || resp == nil {
+		return nil, err
+	}
+
+	// TODO: write a resp.ToMap() function on https://godoc.org/github.com/coreos/etcd/etcdserver/etcdserverpb#RangeResponse
+	result := make(map[string]string)
+	for _, x := range resp.Kvs {
+		result[string(x.Key)] = string(x.Value)
+	}
+	return result, nil
+}
+
+// Txn runs a transaction on the client connection. This has the same signature
+// as our EmbdEtcd Txn function.
+func (obj *ClientEtcd) Txn(ifcmps []etcd.Cmp, thenops, elseops []etcd.Op) (*etcd.TxnResponse, error) {
+	return obj.client.KV.Txn(context.TODO()).If(ifcmps...).Then(thenops...).Else(elseops...).Commit()
+}

etcd/deploy.go

@@ -0,0 +1,171 @@
+// Mgmt
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
+// Written by James Shubin <james@shubin.ca> and the project contributors
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+package etcd
+
+import (
+	"fmt"
+	"strconv"
+	"strings"
+
+	etcd "github.com/coreos/etcd/clientv3"
+	errwrap "github.com/pkg/errors"
+)
+
+const (
+	deployPath  = "deploy"
+	payloadPath = "payload"
+	hashPath    = "hash"
+)
+
+// WatchDeploy returns a channel which spits out events on new deploy activity.
+// FIXME: It should close the channel when it's done, and spit out errors when
+// something goes wrong.
+func WatchDeploy(obj *EmbdEtcd) chan error {
+	// key structure is $NS/deploy/$id/payload = $data
+	path := fmt.Sprintf("%s/%s/", NS, deployPath)
+	ch := make(chan error, 1)
+	// FIXME: fix our API so that we get a close event on shutdown.
+	callback := func(re *RE) error {
+		// TODO: is this even needed? it used to happen on conn errors
+		//log.Printf("Etcd: Watch: Path: %v", path) // event
+		if re == nil || re.response.Canceled {
+			return fmt.Errorf("watch is empty") // will cause a CtxError+retry
+		}
+		if len(ch) == 0 { // send event only if one isn't pending
+			ch <- nil // event
+		}
+		return nil
+	}
+	_, _ = obj.AddWatcher(path, callback, true, false, etcd.WithPrefix()) // no need to check errors
+	return ch
+}
+
+// GetDeploys gets all the available deploys.
+func GetDeploys(obj Client) (map[uint64]string, error) {
+	// key structure is $NS/deploy/$id/payload = $data
+	path := fmt.Sprintf("%s/%s/", NS, deployPath)
+	keyMap, err := obj.Get(path, etcd.WithPrefix(), etcd.WithSort(etcd.SortByKey, etcd.SortAscend))
+	if err != nil {
+		return nil, errwrap.Wrapf(err, "could not get deploy")
+	}
+	result := make(map[uint64]string)
+	for key, val := range keyMap {
+		if !strings.HasPrefix(key, path) { // sanity check
+			continue
+		}
+
+		str := strings.Split(key[len(path):], "/")
+		if len(str) != 2 {
+			return nil, fmt.Errorf("unexpected chunk count of %d", len(str))
+		}
+		if s := str[1]; s != payloadPath {
+			continue // skip, maybe there are other future additions
+		}
+
+		var id uint64
+		var err error
+		x := str[0]
+		if id, err = strconv.ParseUint(x, 10, 64); err != nil {
+			return nil, fmt.Errorf("invalid id of `%s`", x)
+		}
+
+		// TODO: do some sort of filtering here?
+		//log.Printf("Etcd: GetDeploys(%s): Id => Data: %d => %s", key, id, val)
+		result[id] = val
+	}
+	return result, nil
+}
+
+// GetDeploy gets the latest deploy if id == 0, otherwise it returns the deploy
+// with the specified id if it exists.
+// FIXME: implement this more efficiently so that it doesn't have to download *all* the old deploys from etcd!
+func GetDeploy(obj Client, id uint64) (string, error) {
+	result, err := GetDeploys(obj)
+	if err != nil {
+		return "", err
+	}
+	if id != 0 {
+		str, exists := result[id]
+		if !exists {
+			return "", fmt.Errorf("can't find id `%d`", id)
+		}
+		return str, nil
+	}
+	// find the latest id
+	var max uint64
+	for i := range result {
+		if i > max {
+			max = i
+		}
+	}
+	if max == 0 {
+		return "", nil // no results yet
+	}
+	return result[max], nil
+}
+
+// AddDeploy adds a new deploy. It takes an id and ensures it's sequential. If
+// hash is not empty, then it will check that the pHash matches what the
+// previous hash was, and also adds this new hash along side the id. This is
+// useful to make sure you get a linear chain of git patches, and to avoid two
+// contributors pushing conflicting deploys. This isn't git specific, and so any
+// arbitrary string hash can be used.
+// FIXME: prune old deploys from the store when they aren't needed anymore...
+func AddDeploy(obj Client, id uint64, hash, pHash string, data *string) error {
+	// key structure is $NS/deploy/$id/payload = $data
+	// key structure is $NS/deploy/$id/hash = $hash
+	path := fmt.Sprintf("%s/%s/%d/%s", NS, deployPath, id, payloadPath)
+	tPath := fmt.Sprintf("%s/%s/%d/%s", NS, deployPath, id, hashPath)
+	ifs := []etcd.Cmp{} // list matching the desired state
+	ops := []etcd.Op{}  // list of ops in this transaction (then)
+
+	// TODO: use https://github.com/coreos/etcd/pull/7417 if merged
+	// we're append only, so ensure this unique deploy id doesn't exist
+	ifs = append(ifs, etcd.Compare(etcd.Version(path), "=", 0)) // KeyMissing
+	//ifs = append(ifs, etcd.KeyMissing(path))
+
+	// don't look for previous deploy if this is the first deploy ever
+	if id > 1 {
+		// we append sequentially, so ensure previous key *does* exist
+		prev := fmt.Sprintf("%s/%s/%d/%s", NS, deployPath, id-1, payloadPath)
+		ifs = append(ifs, etcd.Compare(etcd.Version(prev), ">", 0)) // KeyExists
+		//ifs = append(ifs, etcd.KeyExists(prev))
+
+		if hash != "" && pHash != "" {
+			// does the previously stored hash match what we expect?
+			prevHash := fmt.Sprintf("%s/%s/%d/%s", NS, deployPath, id-1, hashPath)
+			ifs = append(ifs, etcd.Compare(etcd.Value(prevHash), "=", pHash))
+		}
+	}
+
+	ops = append(ops, etcd.OpPut(path, *data))
+	if hash != "" {
+		ops = append(ops, etcd.OpPut(tPath, hash)) // store new hash as well
+	}
+
+	// it's important to do this in one transaction, and atomically, because
+	// this way, we only generate one watch event, and only when it's needed
+	result, err := obj.Txn(ifs, ops, nil)
+	if err != nil {
+		return errwrap.Wrapf(err, "error creating deploy id %d: %s", id)
+	}
+	if !result.Succeeded {
+		return fmt.Errorf("could not create deploy id %d", id)
+	}
+	return nil // success
+}

etcd/etcd.go

@@ -1,5 +1,5 @@
 // Mgmt
-// Copyright (C) 2013-2017+ James Shubin and the project contributors
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
 // Written by James Shubin <james@shubin.ca> and the project contributors
 //
 // This program is free software: you can redistribute it and/or modify
@@ -79,16 +79,23 @@ import (
 
 // constant parameters which may need to be tweaked or customized
 const (
-	NS                      = "_mgmt" // root namespace for mgmt operations
-	seedSentinel            = "_seed" // you must not name your hostname this
-	MaxStartServerTimeout   = 60      // max number of seconds to wait for server to start
-	MaxStartServerRetries   = 3       // number of times to retry starting the etcd server
-	maxClientConnectRetries = 5       // number of times to retry consecutive connect failures
-	selfRemoveTimeout       = 3       // give unnominated members a chance to self exit
-	exitDelay               = 3       // number of sec of inactivity after exit to clean up
-	DefaultIdealClusterSize = 5       // default ideal cluster size target for initial seed
+	NS                      = "/_mgmt" // root namespace for mgmt operations
+	seedSentinel            = "_seed"  // you must not name your hostname this
+	MaxStartServerTimeout   = 60       // max number of seconds to wait for server to start
+	MaxStartServerRetries   = 3        // number of times to retry starting the etcd server
+	maxClientConnectRetries = 5        // number of times to retry consecutive connect failures
+	selfRemoveTimeout       = 3        // give unnominated members a chance to self exit
+	exitDelay               = 3        // number of sec of inactivity after exit to clean up
+	DefaultIdealClusterSize = 5        // default ideal cluster size target for initial seed
 	DefaultClientURL        = "127.0.0.1:2379"
 	DefaultServerURL        = "127.0.0.1:2380"
+	// DefaultMaxTxnOps is the maximum number of operations to run in a
+	// single etcd transaction. If you exceed this limit, it is possible
+	// that you have either an extremely large code base, or that you have
+	// some code which is possibly not as efficient as it could be. Let us
+	// know so that we can analyze the situation, and increase this if
+	// necessary.
+	DefaultMaxTxnOps = 512
 )
 
 var (
@@ -170,18 +177,21 @@ type EmbdEtcd struct { // EMBeddeD etcd
 	ctxErr error // permanent ctx error
 
 	// exit and cleanup related
-	cancelLock  sync.Mutex // lock for the cancels list
-	cancels     []func()   // array of every cancel function for watches
-	exiting     bool
-	exitchan    chan struct{}
-	exitTimeout <-chan time.Time
+	cancelLock sync.Mutex // lock for the cancels list
+	cancels    []func()   // array of every cancel function for watches
+	exiting    bool
+	exitchan   chan struct{}
+	exitchanCb chan struct{}
+	exitwg     *sync.WaitGroup // wait for main loops to shutdown
 
-	hostname   string
-	memberID   uint64            // cluster membership id of server if running
-	endpoints  etcdtypes.URLsMap // map of servers a client could connect to
-	clientURLs etcdtypes.URLs    // locations to listen for clients if i am a server
-	serverURLs etcdtypes.URLs    // locations to listen for servers if i am a server (peer)
-	noServer   bool              // disable all server peering if true
+	hostname            string
+	memberID            uint64            // cluster membership id of server if running
+	endpoints           etcdtypes.URLsMap // map of servers a client could connect to
+	clientURLs          etcdtypes.URLs    // locations to listen for clients if i am a server
+	serverURLs          etcdtypes.URLs    // locations to listen for servers if i am a server (peer)
+	advertiseClientURLs etcdtypes.URLs    // client urls to advertise
+	advertiseServerURLs etcdtypes.URLs    // server urls to advertise
+	noServer            bool              // disable all server peering if true
 
 	// local tracked state
 	nominated        etcdtypes.URLsMap // copy of who's nominated to locally track state
@@ -208,32 +218,39 @@ type EmbdEtcd struct { // EMBeddeD etcd
 }
 
 // NewEmbdEtcd creates the top level embedded etcd struct client and server obj.
-func NewEmbdEtcd(hostname string, seeds, clientURLs, serverURLs etcdtypes.URLs, noServer bool, idealClusterSize uint16, flags Flags, prefix string, converger converger.Converger) *EmbdEtcd {
+func NewEmbdEtcd(hostname string, seeds, clientURLs, serverURLs, advertiseClientURLs, advertiseServerURLs etcdtypes.URLs, noServer bool, idealClusterSize uint16, flags Flags, prefix string, converger converger.Converger) *EmbdEtcd {
 	endpoints := make(etcdtypes.URLsMap)
 	if hostname == seedSentinel { // safety
 		return nil
 	}
+	if noServer && len(seeds) == 0 {
+		log.Printf("Etcd: need at least one seed if running with --no-server!")
+		return nil
+	}
 	if len(seeds) > 0 {
 		endpoints[seedSentinel] = seeds
 		idealClusterSize = 0 // unset, get from running cluster
 	}
 	obj := &EmbdEtcd{
-		exitchan:    make(chan struct{}), // exit signal for main loop
-		exitTimeout: nil,
-		awq:         make(chan *AW),
-		wevents:     make(chan *RE),
-		setq:        make(chan *KV),
-		getq:        make(chan *GQ),
-		delq:        make(chan *DL),
-		txnq:        make(chan *TN),
+		exitchan:   make(chan struct{}), // exit signal for main loop
+		exitchanCb: make(chan struct{}),
+		exitwg:     &sync.WaitGroup{},
+		awq:        make(chan *AW),
+		wevents:    make(chan *RE),
+		setq:       make(chan *KV),
+		getq:       make(chan *GQ),
+		delq:       make(chan *DL),
+		txnq:       make(chan *TN),
 
 		nominated: make(etcdtypes.URLsMap),
 
-		hostname:   hostname,
-		endpoints:  endpoints,
-		clientURLs: clientURLs,
-		serverURLs: serverURLs,
-		noServer:   noServer,
+		hostname:            hostname,
+		endpoints:           endpoints,
+		clientURLs:          clientURLs,
+		serverURLs:          serverURLs,
+		advertiseClientURLs: advertiseClientURLs,
+		advertiseServerURLs: advertiseServerURLs,
+		noServer:            noServer,
 
 		idealClusterSize: idealClusterSize,
 		converger:        converger,
@@ -261,6 +278,11 @@ func NewEmbdEtcd(hostname string, seeds, clientURLs, serverURLs etcdtypes.URLs,
 	return obj
 }
 
+// GetClient returns a handle to the raw etcd client object for those scenarios.
+func (obj *EmbdEtcd) GetClient() *etcd.Client {
+	return obj.client
+}
+
 // GetConfig returns the config struct to be used for the etcd client connect.
 func (obj *EmbdEtcd) GetConfig() etcd.Config {
 	endpoints := []string{}
@@ -317,7 +339,7 @@ func (obj *EmbdEtcd) Connect(reconnect bool) error {
 			emax++
 			if emax > maxClientConnectRetries {
 				log.Printf("Etcd: The dataDir (%s) might be inconsistent or corrupt.", obj.dataDir)
-				log.Printf("Etcd: Please see: %s", "https://github.com/purpleidea/mgmt/blob/master/DOCUMENTATION.md#what-does-the-error-message-about-an-inconsistent-datadir-mean")
+				log.Printf("Etcd: Please see: %s", "https://github.com/purpleidea/mgmt/blob/master/docs/faq.md#what-does-the-error-message-about-an-inconsistent-datadir-mean")
 				obj.cError = fmt.Errorf("can't find an available endpoint")
 				return obj.cError
 			}
@@ -359,11 +381,11 @@ func (obj *EmbdEtcd) Startup() error {
 	go obj.Loop()   // start main loop
 
 	// TODO: implement native etcd watcher method on member API changes
-	path := fmt.Sprintf("/%s/nominated/", NS)
+	path := fmt.Sprintf("%s/nominated/", NS)
 	go obj.AddWatcher(path, obj.nominateCallback, true, false, etcd.WithPrefix()) // no block
 
 	// setup ideal cluster size watcher
-	key := fmt.Sprintf("/%s/idealClusterSize", NS)
+	key := fmt.Sprintf("%s/idealClusterSize", NS)
 	go obj.AddWatcher(key, obj.idealClusterSizeCallback, true, false) // no block
 
 	// if we have no endpoints, it means we are bootstrapping...
@@ -389,16 +411,20 @@ func (obj *EmbdEtcd) Startup() error {
 	}
 
 	if !obj.noServer {
-		path := fmt.Sprintf("/%s/volunteers/", NS)
+		path := fmt.Sprintf("%s/volunteers/", NS)
 		go obj.AddWatcher(path, obj.volunteerCallback, true, false, etcd.WithPrefix()) // no block
 	}
 
 	// if i am alone and will have to be a server...
 	if !obj.noServer && bootstrapping {
 		log.Printf("Etcd: Bootstrapping...")
+		surls := obj.serverURLs
+		if len(obj.advertiseServerURLs) > 0 {
+			surls = obj.advertiseServerURLs
+		}
 		// give an initial value to the obj.nominate map we keep in sync
 		// this emulates Nominate(obj, obj.hostname, obj.serverURLs)
-		obj.nominated[obj.hostname] = obj.serverURLs // initial value
+		obj.nominated[obj.hostname] = surls // initial value
 		// NOTE: when we are stuck waiting for the server to start up,
 		// it is probably happening on this call right here...
 		obj.nominateCallback(nil) // kick this off once
@@ -407,8 +433,12 @@ func (obj *EmbdEtcd) Startup() error {
 	// self volunteer
 	if !obj.noServer && len(obj.serverURLs) > 0 {
 		// we run this in a go routine because it blocks waiting for server
+		surls := obj.serverURLs
+		if len(obj.advertiseServerURLs) > 0 {
+			surls = obj.advertiseServerURLs
+		}
 		log.Printf("Etcd: Startup: Volunteering...")
-		go Volunteer(obj, obj.serverURLs)
+		go Volunteer(obj, surls)
 	}
 
 	if bootstrapping {
@@ -419,7 +449,7 @@ func (obj *EmbdEtcd) Startup() error {
 		}
 	}
 
-	go obj.AddWatcher(fmt.Sprintf("/%s/endpoints/", NS), obj.endpointCallback, true, false, etcd.WithPrefix())
+	go obj.AddWatcher(fmt.Sprintf("%s/endpoints/", NS), obj.endpointCallback, true, false, etcd.WithPrefix())
 
 	if err := obj.Connect(false); err != nil { // don't exit from this Startup function until connected!
 		return err
@@ -449,7 +479,8 @@ func (obj *EmbdEtcd) Destroy() error {
 	}
 	obj.cancelLock.Unlock()
 
-	obj.exitchan <- struct{}{} // cause main loop to exit
+	close(obj.exitchan) // cause main loop to exit
+	close(obj.exitchanCb)
 
 	obj.rLock.Lock()
 	if obj.client != nil {
@@ -462,6 +493,7 @@ func (obj *EmbdEtcd) Destroy() error {
 	//if obj.server != nil {
 	//	return obj.DestroyServer()
 	//}
+	obj.exitwg.Wait()
 	return nil
 }
 
@@ -537,7 +569,7 @@ func (obj *EmbdEtcd) CtxError(ctx context.Context, err error) (context.Context,
 		// tmin <= texp^iter - 1 <= tmax // TODO: check my math
 		return time.Duration(math.Min(math.Max(math.Pow(float64(texp), float64(iter))-1.0, float64(tmin)), float64(tmax))) * time.Millisecond
 	}
-	var isTimeout = false
+	var isTimeout bool
 	var iter int // = 0
 	if ctxerr, ok := ctx.Value(ctxErr).(error); ok {
 		if obj.flags.Debug {
@@ -703,12 +735,15 @@ func (obj *EmbdEtcd) CtxError(ctx context.Context, err error) (context.Context,
 
 // CbLoop is the loop where callback execution is serialized.
 func (obj *EmbdEtcd) CbLoop() {
+	obj.exitwg.Add(1)
+	defer obj.exitwg.Done()
 	cuid := obj.converger.Register()
 	cuid.SetName("Etcd: CbLoop")
 	defer cuid.Unregister()
 	if e := obj.Connect(false); e != nil {
 		return // fatal
 	}
+	var exitTimeout <-chan time.Time // = nil is implied
 	// we use this timer because when we ignore un-converge events and loop,
 	// we reset the ConvergedTimer case statement, ruining the timeout math!
 	cuid.StartTimer()
@@ -748,8 +783,18 @@ func (obj *EmbdEtcd) CbLoop() {
 				log.Printf("Trace: Etcd: CbLoop: Event: FinishLoop")
 			}
 
+		// exit loop signal
+		case <-obj.exitchanCb:
+			obj.exitchanCb = nil
+			log.Println("Etcd: Exiting loop shortly...")
+			// activate exitTimeout switch which only opens after N
+			// seconds of inactivity in this select switch, which
+			// lets everything get bled dry to avoid blocking calls
+			// which would otherwise block us from exiting cleanly!
+			exitTimeout = util.TimeAfterOrBlock(exitDelay)
+
 		// exit loop commit
-		case <-obj.exitTimeout:
+		case <-exitTimeout:
 			log.Println("Etcd: Exiting callback loop!")
 			cuid.StopTimer() // clean up nicely
 			return
@@ -759,12 +804,15 @@ func (obj *EmbdEtcd) CbLoop() {
 
 // Loop is the main loop where everything is serialized.
 func (obj *EmbdEtcd) Loop() {
+	obj.exitwg.Add(1) // TODO: add these to other go routines?
+	defer obj.exitwg.Done()
 	cuid := obj.converger.Register()
 	cuid.SetName("Etcd: Loop")
 	defer cuid.Unregister()
 	if e := obj.Connect(false); e != nil {
 		return // fatal
 	}
+	var exitTimeout <-chan time.Time // = nil is implied
 	cuid.StartTimer()
 	for {
 		ctx := context.Background() // TODO: inherit as input argument?
@@ -899,15 +947,16 @@ func (obj *EmbdEtcd) Loop() {
 
 		// exit loop signal
 		case <-obj.exitchan:
+			obj.exitchan = nil
 			log.Println("Etcd: Exiting loop shortly...")
 			// activate exitTimeout switch which only opens after N
 			// seconds of inactivity in this select switch, which
 			// lets everything get bled dry to avoid blocking calls
 			// which would otherwise block us from exiting cleanly!
-			obj.exitTimeout = util.TimeAfterOrBlock(exitDelay)
+			exitTimeout = util.TimeAfterOrBlock(exitDelay)
 
 		// exit loop commit
-		case <-obj.exitTimeout:
+		case <-exitTimeout:
 			log.Println("Etcd: Exiting loop!")
 			cuid.StopTimer() // clean up nicely
 			return
@@ -1436,14 +1485,21 @@ func (obj *EmbdEtcd) nominateCallback(re *RE) error {
 			// client connects to one of the obj.endpoints servers...
 			log.Printf("Etcd: Addresses are: %s", addresses)
 
+			surls := obj.serverURLs
+			if len(obj.advertiseServerURLs) > 0 {
+				surls = obj.advertiseServerURLs
+			}
 			// XXX: just put this wherever for now so we don't block
 			// nominate self so "member" list is correct for peers to see
-			Nominate(obj, obj.hostname, obj.serverURLs)
+			Nominate(obj, obj.hostname, surls)
 			// XXX: if this fails, where will we retry this part ?
 		}
 
 		// advertise client urls
 		if curls := obj.clientURLs; len(curls) > 0 {
+			if len(obj.advertiseClientURLs) > 0 {
+				curls = obj.advertiseClientURLs
+			}
 			// XXX: don't advertise local addresses! 127.0.0.1:2381 doesn't really help remote hosts
 			// XXX: but sometimes this is what we want... hmmm how do we decide? filter on callback?
 			AdvertiseEndpoints(obj, curls)
@@ -1578,7 +1634,7 @@ func (obj *EmbdEtcd) idealClusterSizeCallback(re *RE) error {
 		log.Printf("Trace: Etcd: idealClusterSizeCallback()")
 		defer log.Printf("Trace: Etcd: idealClusterSizeCallback(): Finished!")
 	}
-	path := fmt.Sprintf("/%s/idealClusterSize", NS)
+	path := fmt.Sprintf("%s/idealClusterSize", NS)
 	for _, event := range re.response.Events {
 		if key := bytes.NewBuffer(event.Kv.Key).String(); key != path {
 			continue
@@ -1647,15 +1703,25 @@ func (obj *EmbdEtcd) StartServer(newCluster bool, peerURLsMap etcdtypes.URLsMap)
 		initialPeerURLsMap[memberName] = peerURLs
 	}
 
+	aCUrls := obj.clientURLs
+	if len(obj.advertiseClientURLs) > 0 {
+		aCUrls = obj.advertiseClientURLs
+	}
+	aPUrls := peerURLs
+	if len(obj.advertiseServerURLs) > 0 {
+		aPUrls = obj.advertiseServerURLs
+	}
+
 	// embed etcd
 	cfg := embed.NewConfig()
 	cfg.Name = memberName // hostname
 	cfg.Dir = obj.dataDir
-	cfg.ACUrls = obj.clientURLs
-	cfg.APUrls = peerURLs
 	cfg.LCUrls = obj.clientURLs
 	cfg.LPUrls = peerURLs
+	cfg.ACUrls = aCUrls
+	cfg.APUrls = aPUrls
 	cfg.StrictReconfigCheck = false // XXX: workaround https://github.com/coreos/etcd/issues/6305
+	cfg.MaxTxnOps = DefaultMaxTxnOps
 
 	cfg.InitialCluster = initialPeerURLsMap.String() // including myself!
 	if newCluster {

etcd/etcd_test.go

@@ -0,0 +1,49 @@
+// Mgmt
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
+// Written by James Shubin <james@shubin.ca> and the project contributors
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+package etcd
+
+import (
+	"testing"
+
+	etcdtypes "github.com/coreos/etcd/pkg/types"
+)
+
+func TestNewEmbdEtcd(t *testing.T) {
+	// should return a new etcd object
+
+	noServer := false
+	var flags Flags
+
+	obj := NewEmbdEtcd("", nil, nil, nil, nil, nil, noServer, 0, flags, "", nil)
+	if obj == nil {
+		t.Fatal("failed to create server object")
+	}
+}
+
+func TestNewEmbdEtcdConfigValidation(t *testing.T) {
+	// running --no-server with no --seeds specified should fail early
+
+	seeds := make(etcdtypes.URLs, 0)
+	noServer := true
+	var flags Flags
+
+	obj := NewEmbdEtcd("", seeds, nil, nil, nil, nil, noServer, 0, flags, "", nil)
+	if obj != nil {
+		t.Fatal("server initialization should fail on invalid configuration")
+	}
+}

etcd/fs/file.go

@@ -0,0 +1,540 @@
+// Mgmt
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
+// Written by James Shubin <james@shubin.ca> and the project contributors
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+package fs
+
+import (
+	"bytes"
+	"encoding/gob"
+	"fmt"
+	"io"
+	"log"
+	"os"
+	"path"
+	"strings"
+	"syscall"
+	"time"
+
+	etcd "github.com/coreos/etcd/clientv3" // "clientv3"
+	errwrap "github.com/pkg/errors"
+)
+
+func init() {
+	gob.Register(&File{})
+}
+
+// File represents a file node. This is the node of our tree structure. This is
+// not thread safe, and you can have at most one open file handle at a time.
+type File struct {
+	// FIXME: add a rwmutex to make this thread safe
+	fs *Fs // pointer to file system
+
+	Path    string // relative path to file, trailing slash if it's a directory
+	Mode    os.FileMode
+	ModTime time.Time
+	//Size int64 // XXX: cache the size to avoid full file downloads for stat!
+
+	Children []*File // dir's use this
+	Hash     string  // string not []byte so it's readable, matches data
+
+	data      []byte // cache of the data. private so it doesn't get encoded
+	cursor    int64
+	dirCursor int64
+
+	readOnly bool // is the file read-only?
+	closed   bool // is file closed?
+}
+
+// path returns the expected path to the actual file in etcd.
+func (obj *File) path() string {
+	// keys are prefixed with the hash-type eg: {sha256} to allow different
+	// superblocks to share the same data prefix even with different hashes
+	return fmt.Sprintf("%s/{%s}%s", obj.fs.sb.DataPrefix, obj.fs.Hash, obj.Hash)
+}
+
+// cache downloads the file contents from etcd and stores them in our cache.
+func (obj *File) cache() error {
+	if obj.Mode.IsDir() {
+		return nil
+	}
+
+	h, err := obj.fs.hash(obj.data) // update hash
+	if err != nil {
+		return err
+	}
+	if h == obj.Hash { // we already have the correct data cached
+		return nil
+	}
+
+	p := obj.path() // get file data from this path in etcd
+
+	result, err := obj.fs.get(p) // download the file...
+	if err != nil {
+		return err
+	}
+	if result == nil || len(result) == 0 { // nothing found
+		return err
+	}
+	data, exists := result[p]
+	if !exists {
+		return fmt.Errorf("could not find data") // programming error?
+	}
+	obj.data = data // save
+	return nil
+}
+
+// findNode is the "in array" equivalent for searching through a dir's children.
+// You must *not* specify an absolute path as the search string, but rather you
+// should specify the name. To search for something name "bar" inside a dir
+// named "/tmp/foo/", you just pass in "bar", not "/tmp/foo/bar".
+func (obj *File) findNode(name string) (*File, bool) {
+	for _, node := range obj.Children {
+		if name == node.Path {
+			return node, true // found
+		}
+	}
+	return nil, false // not found
+}
+
+func fileCreate(fs *Fs, name string) (*File, error) {
+	if name == "" {
+		return nil, fmt.Errorf("invalid input path")
+	}
+	if !strings.HasPrefix(name, "/") {
+		return nil, fmt.Errorf("invalid input path (not absolute)")
+	}
+	cleanPath := path.Clean(name) // remove possible trailing slashes
+
+	// try to add node to tree by first finding the parent node
+	parentPath, filePath := path.Split(cleanPath) // looking for this
+
+	node, err := fs.find(parentPath)
+	if err != nil { // might be ErrNotExist
+		return nil, err
+	}
+
+	fi, err := node.Stat()
+	if err != nil {
+		return nil, err
+	}
+	if !fi.IsDir() { // is the parent a suitable home?
+		return nil, &os.PathError{Op: "create", Path: name, Err: syscall.ENOTDIR}
+	}
+
+	f, exists := node.findNode(filePath) // does file already exist inside?
+	if exists {                          // already exists, overwrite!
+		if err := f.Truncate(0); err != nil {
+			return nil, err
+		}
+		return f, nil
+	}
+
+	data := []byte("")      // empty file contents
+	h, err := fs.hash(data) // TODO: use memoized value?
+	if err != nil {
+		return &File{}, err // TODO: nil instead?
+	}
+
+	f = &File{
+		fs:   fs,
+		Path: filePath, // the relative path chunk (not incl. dir name)
+		Hash: h,
+		data: data,
+	}
+
+	// add to parent
+	node.Children = append(node.Children, f)
+
+	// push new file up if not on server, and then push up the metadata
+	if err := f.Sync(); err != nil {
+		return f, err // TODO: ok to return the file so user can run sync?
+	}
+
+	return f, nil
+}
+
+func fileOpen(fs *Fs, name string) (*File, error) {
+	if name == "" {
+		return nil, fmt.Errorf("invalid input path")
+	}
+	if !strings.HasPrefix(name, "/") {
+		return nil, fmt.Errorf("invalid input path (not absolute)")
+	}
+	cleanPath := path.Clean(name) // remove possible trailing slashes
+
+	node, err := fs.find(cleanPath)
+	if err != nil { // might be ErrNotExist
+		return &File{}, err // TODO: nil instead?
+	}
+
+	// download file contents into obj.data
+	if err := node.cache(); err != nil {
+		return &File{}, err // TODO: nil instead?
+	}
+
+	//fi, err := node.Stat()
+	//if err != nil {
+	//	return nil, err
+	//}
+	//if fi.IsDir() { // can we open a directory? - yes we can apparently
+	//	return nil, fmt.Errorf("file is a directory")
+	//}
+
+	node.readOnly = true // as per docs, fileOpen opens files as read-only
+	node.closed = false  // as per docs, fileOpen opens files as read-only
+
+	return node, nil
+}
+
+// Close closes the file handle. This will try and run Sync automatically.
+func (obj *File) Close() error {
+	if !obj.readOnly {
+		obj.ModTime = time.Now()
+	}
+
+	if err := obj.Sync(); err != nil {
+		return err
+	}
+
+	// FIXME: there is a big implementation mistake between the metadata
+	// node and the file handle, since they're currently sharing a struct!
+
+	// invalidate all of the fields
+	//obj.fs = nil
+
+	//obj.Path = ""
+	//obj.Mode = os.FileMode(0)
+	//obj.ModTime = time.Time{}
+
+	//obj.Children = nil
+	//obj.Hash = ""
+
+	//obj.data = nil
+	obj.cursor = 0
+	obj.readOnly = false
+
+	obj.closed = true
+	return nil
+}
+
+// Name returns the path of the file.
+func (obj *File) Name() string {
+	return obj.Path
+}
+
+// Stat returns some information about the file.
+func (obj *File) Stat() (os.FileInfo, error) {
+	// download file contents into obj.data
+	if err := obj.cache(); err != nil { // needed so Size() works correctly
+		return nil, err
+	}
+	return &FileInfo{ // everything is actually stored in the main file node
+		file: obj,
+	}, nil
+}
+
+// Sync flushes the file contents to the server and calls the filesystem
+// metadata sync as well.
+// FIXME: instead of a txn, run a get and then a put in two separate stages. if
+// the get already found the data up there, then we don't need to push it all in
+// the put phase. with the txn it is always all sent up even if the put is never
+// needed. the get should just be a "key exists" test, and not a download of the
+// whole file. if we *do* do the download, we can byte-by-byte check for hash
+// collisions and panic if we find one :)
+func (obj *File) Sync() error {
+	if obj.closed {
+		return ErrFileClosed
+	}
+
+	p := obj.path() // store file data at this path in etcd
+
+	// TODO: use https://github.com/coreos/etcd/pull/7417 if merged
+	cmp := etcd.Compare(etcd.Version(p), "=", 0) // KeyMissing
+	//cmp := etcd.KeyMissing(p))
+
+	op := etcd.OpPut(p, string(obj.data)) // this pushes contents to server
+
+	// it's important to do this in one transaction, and atomically, because
+	// this way, we only generate one watch event, and only when it's needed
+	result, err := obj.fs.txn([]etcd.Cmp{cmp}, []etcd.Op{op}, nil)
+	if err != nil {
+		return errwrap.Wrapf(err, "sync error with: %s (%s)", obj.Path, p)
+	}
+	if !result.Succeeded {
+		if obj.fs.Debug {
+			log.Printf("debug: data already exists in storage")
+		}
+	}
+
+	if err := obj.fs.sync(); err != nil { // push metadata up to server
+		return err
+	}
+	return nil
+}
+
+// Truncate trims the file to the requested size. Since our file system can only
+// read and write data, but never edit existing data blocks, doing this will not
+// cause more space to be available.
+func (obj *File) Truncate(size int64) error {
+	if obj.closed {
+		return ErrFileClosed
+	}
+	if obj.readOnly {
+		return &os.PathError{Op: "truncate", Path: obj.Path, Err: ErrFileReadOnly}
+	}
+	if size < 0 {
+		return ErrOutOfRange
+	}
+
+	if size > 0 { // if size == 0, we don't need to run cache!
+		// download file contents into obj.data
+		if err := obj.cache(); err != nil {
+			return err
+		}
+	}
+
+	if size > int64(len(obj.data)) {
+		diff := size - int64(len(obj.data))
+		obj.data = append(obj.data, bytes.Repeat([]byte{00}, int(diff))...)
+	} else {
+		obj.data = obj.data[0:size]
+	}
+
+	h, err := obj.fs.hash(obj.data) // update hash
+	if err != nil {
+		return err
+	}
+	obj.Hash = h
+	obj.ModTime = time.Now()
+
+	// this pushes the new data and metadata up to etcd
+	return obj.Sync()
+}
+
+// Read reads up to len(b) bytes from the File. It returns the number of bytes
+// read and any error encountered. At end of file, Read returns 0, io.EOF.
+// NOTE: This reads into the byte input. It's a side effect!
+func (obj *File) Read(b []byte) (n int, err error) {
+	if obj.closed {
+		return 0, ErrFileClosed
+	}
+	if obj.Mode.IsDir() {
+		return 0, fmt.Errorf("file is a directory")
+	}
+
+	// download file contents into obj.data
+	if err := obj.cache(); err != nil {
+		return 0, err // TODO: -1 ?
+	}
+
+	// TODO: can we optimize by reading just the length from etcd, and also
+	// by only downloading the data range we're interested in?
+	if len(b) > 0 && int(obj.cursor) == len(obj.data) {
+		return 0, io.EOF
+	}
+	if len(obj.data)-int(obj.cursor) >= len(b) {
+		n = len(b)
+	} else {
+		n = len(obj.data) - int(obj.cursor)
+	}
+	copy(b, obj.data[obj.cursor:obj.cursor+int64(n)]) // store into input b
+	obj.cursor = obj.cursor + int64(n)                // update cursor
+
+	return
+}
+
+// ReadAt reads len(b) bytes from the File starting at byte offset off. It
+// returns the number of bytes read and the error, if any. ReadAt always returns
+// a non-nil error when n < len(b). At end of file, that error is io.EOF.
+func (obj *File) ReadAt(b []byte, off int64) (n int, err error) {
+	obj.cursor = off
+	return obj.Read(b)
+}
+
+// Readdir lists the contents of the directory and returns a list of file info
+// objects for each entry.
+func (obj *File) Readdir(count int) ([]os.FileInfo, error) {
+	if !obj.Mode.IsDir() {
+		return nil, &os.PathError{Op: "readdir", Path: obj.Name(), Err: syscall.ENOTDIR}
+	}
+
+	children := obj.Children[obj.dirCursor:] // available children to output
+	var l = int64(len(children))             // initially assume to return them all
+	var err error
+
+	// for count > 0, if we return the last entry, also return io.EOF
+	if count > 0 {
+		l = int64(count) // initial assumption
+		if c := len(children); count >= c {
+			l = int64(c)
+			err = io.EOF // this result includes the last dir entry
+		}
+	}
+	obj.dirCursor += l // store our progress
+
+	output := make([]os.FileInfo, l)
+	// TODO: should this be sorted by "directory order" what does that mean?
+	// from `man 3 readdir`: "unlikely that the names will be sorted"
+	for i := range output {
+		output[i] = &FileInfo{
+			file: children[i],
+		}
+	}
+
+	// we're seen the whole directory, so reset the cursor
+	if err == io.EOF || count <= 0 {
+		obj.dirCursor = 0 // TODO: is it okay to reset the cursor?
+	}
+
+	return output, err
+}
+
+// Readdirnames returns a list of name is the current file handle's directory.
+// TODO: this implementation shares the dirCursor with Readdir, is this okay?
+// TODO: should Readdirnames even use a dirCursor at all?
+func (obj *File) Readdirnames(n int) (names []string, _ error) {
+	fis, err := obj.Readdir(n)
+	if fis != nil {
+		for i, x := range fis {
+			if x != nil {
+				names = append(names, fis[i].Name())
+			}
+		}
+	}
+	return names, err
+}
+
+// Seek sets the offset for the next Read or Write on file to offset,
+// interpreted according to whence: 0 means relative to the origin of the file,
+// 1 means relative to the current offset, and 2 means relative to the end. It
+// returns the new offset and an error, if any. The behavior of Seek on a file
+// opened with O_APPEND is not specified.
+func (obj *File) Seek(offset int64, whence int) (int64, error) {
+	if obj.closed {
+		return 0, ErrFileClosed
+	}
+
+	switch whence {
+	case io.SeekStart: // 0
+		obj.cursor = offset
+	case io.SeekCurrent: // 1
+		obj.cursor += offset
+	case io.SeekEnd: // 2
+		// download file contents into obj.data
+		if err := obj.cache(); err != nil {
+			return 0, err // TODO: -1 ?
+		}
+		obj.cursor = int64(len(obj.data)) + offset
+	}
+	return obj.cursor, nil
+}
+
+// Write writes to the given file.
+func (obj *File) Write(b []byte) (n int, err error) {
+	if obj.closed {
+		return 0, ErrFileClosed
+	}
+	if obj.readOnly {
+		return 0, &os.PathError{Op: "write", Path: obj.Path, Err: ErrFileReadOnly}
+	}
+
+	// download file contents into obj.data
+	if err := obj.cache(); err != nil {
+		return 0, err // TODO: -1 ?
+	}
+
+	// calculate the write
+	n = len(b)
+	cur := obj.cursor
+	diff := cur - int64(len(obj.data))
+
+	var tail []byte
+	if n+int(cur) < len(obj.data) {
+		tail = obj.data[n+int(cur):]
+	}
+
+	if diff > 0 {
+		obj.data = append(bytes.Repeat([]byte{00}, int(diff)), b...)
+		obj.data = append(obj.data, tail...)
+	} else {
+		obj.data = append(obj.data[:cur], b...)
+		obj.data = append(obj.data, tail...)
+	}
+
+	h, err := obj.fs.hash(obj.data) // update hash
+	if err != nil {
+		return 0, err // TODO: -1 ?
+	}
+	obj.Hash = h
+	obj.ModTime = time.Now()
+
+	// this pushes the new data and metadata up to etcd
+	if err := obj.Sync(); err != nil {
+		return 0, err // TODO: -1 ?
+	}
+
+	obj.cursor = int64(len(obj.data))
+	return
+}
+
+// WriteAt writes into the given file at a certain offset.
+func (obj *File) WriteAt(b []byte, off int64) (n int, err error) {
+	obj.cursor = off
+	return obj.Write(b)
+}
+
+// WriteString writes a string to the file.
+func (obj *File) WriteString(s string) (n int, err error) {
+	return obj.Write([]byte(s))
+}
+
+// FileInfo is a struct which provides some information about a file handle.
+type FileInfo struct {
+	file *File // anonymous pointer to the actual file
+}
+
+// Name returns the base name of the file.
+func (obj *FileInfo) Name() string {
+	return obj.file.Name()
+}
+
+// Size returns the length in bytes.
+func (obj *FileInfo) Size() int64 {
+	return int64(len(obj.file.data))
+}
+
+// Mode returns the file mode bits.
+func (obj *FileInfo) Mode() os.FileMode {
+	return obj.file.Mode
+}
+
+// ModTime returns the modification time.
+func (obj *FileInfo) ModTime() time.Time {
+	return obj.file.ModTime
+}
+
+// IsDir is an abbreviation for Mode().IsDir().
+func (obj *FileInfo) IsDir() bool {
+	//return obj.file.Mode&os.ModeDir != 0
+	return obj.file.Mode.IsDir()
+}
+
+// Sys returns the underlying data source (can return nil).
+func (obj *FileInfo) Sys() interface{} {
+	return nil // TODO: should we do something better?
+	//return obj.file.fs // TODO: would this work?
+}

etcd/fs/fs.go

@@ -0,0 +1,821 @@
+// Mgmt
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
+// Written by James Shubin <james@shubin.ca> and the project contributors
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+// Package fs implements a very simple and limited file system on top of etcd.
+package fs
+
+import (
+	"bytes"
+	"crypto/sha256"
+	"encoding/gob"
+	"encoding/hex"
+	"errors"
+	"fmt"
+	"hash"
+	"io"
+	"log"
+	"os"
+	"path"
+	"strings"
+	"syscall"
+	"time"
+
+	etcd "github.com/coreos/etcd/clientv3" // "clientv3"
+	rpctypes "github.com/coreos/etcd/etcdserver/api/v3rpc/rpctypes"
+	errwrap "github.com/pkg/errors"
+	"github.com/spf13/afero"
+	context "golang.org/x/net/context"
+)
+
+func init() {
+	gob.Register(&superBlock{})
+}
+
+const (
+	// EtcdTimeout is the timeout to wait before erroring.
+	EtcdTimeout = 5 * time.Second // FIXME: chosen arbitrarily
+	// DefaultDataPrefix is the default path for data storage in etcd.
+	DefaultDataPrefix = "/_etcdfs/data"
+	// DefaultHash is the default hashing algorithm to use.
+	DefaultHash = "sha256"
+	// PathSeparator is the path separator to use on this filesystem.
+	PathSeparator = os.PathSeparator // usually the slash character
+)
+
+// TODO: https://dave.cheney.net/2016/04/07/constant-errors
+var (
+	IsPathSeparator = os.IsPathSeparator
+
+	// ErrNotImplemented is returned when something is not implemented by design.
+	ErrNotImplemented = errors.New("not implemented")
+
+	// ErrExist is returned when requested path already exists.
+	ErrExist = os.ErrExist
+
+	// ErrNotExist is returned when we can't find the requested path.
+	ErrNotExist = os.ErrNotExist
+
+	ErrFileClosed   = errors.New("File is closed")
+	ErrFileReadOnly = errors.New("File handle is read only")
+	ErrOutOfRange   = errors.New("Out of range")
+)
+
+// Fs is a specialized afero.Fs implementation for etcd. It implements a small
+// subset of the features, and has some special properties. In particular, file
+// data is stored with it's unique reference being a hash of the data. In this
+// way, you cannot actually edit a file, but rather you create a new one, and
+// update the metadata pointer to point to the new blob. This might seem slow,
+// but it has the unique advantage of being relatively straight forward to
+// implement, and repeated uploads of the same file cost almost nothing. Since
+// etcd isn't meant for large file systems, this fits the desired use case.
+// This implementation is designed to have a single writer for each superblock,
+// but as many readers as you like.
+// FIXME: this is not currently thread-safe, nor is it clear if it needs to be.
+// XXX: we probably aren't updating the modification time everywhere we should!
+// XXX: because we never delete data blocks, we need to occasionally "vacuum".
+// XXX: this is harder because we need to list of *all* metadata paths, if we
+// want them to be able to share storage backends. (we do)
+type Fs struct {
+	Client *etcd.Client
+
+	Metadata string // location of "superblock" for this filesystem
+
+	DataPrefix string // prefix of data storage (no trailing slashes)
+	Hash       string // eg: sha256
+
+	Debug bool
+
+	sb      *superBlock
+	mounted bool
+}
+
+// superBlock is the metadata structure of everything stored outside of the data
+// section in etcd. Its fields need to be exported or they won't get marshalled.
+type superBlock struct {
+	DataPrefix string // prefix of data storage
+	Hash       string // hashing algorithm used
+
+	Tree *File // filesystem tree
+}
+
+// NewEtcdFs creates a new filesystem handle on an etcd client connection. You
+// must specify the metadata string that you wish to use.
+func NewEtcdFs(client *etcd.Client, metadata string) afero.Fs {
+	return &Fs{
+		Client:   client,
+		Metadata: metadata,
+	}
+}
+
+// get a number of values from etcd.
+func (obj *Fs) get(path string, opts ...etcd.OpOption) (map[string][]byte, error) {
+	ctx, cancel := context.WithTimeout(context.Background(), EtcdTimeout)
+	resp, err := obj.Client.Get(ctx, path, opts...)
+	cancel()
+	if err != nil || resp == nil {
+		return nil, err
+	}
+
+	// TODO: write a resp.ToMap() function on https://godoc.org/github.com/coreos/etcd/etcdserver/etcdserverpb#RangeResponse
+	result := make(map[string][]byte) // formerly: map[string][]byte
+	for _, x := range resp.Kvs {
+		result[string(x.Key)] = x.Value // formerly: bytes.NewBuffer(x.Value).String()
+	}
+
+	return result, nil
+}
+
+// put a value into etcd.
+func (obj *Fs) put(path string, data []byte, opts ...etcd.OpOption) error {
+	ctx, cancel := context.WithTimeout(context.Background(), EtcdTimeout)
+	_, err := obj.Client.Put(ctx, path, string(data), opts...) // TODO: obj.Client.KV ?
+	cancel()
+	if err != nil {
+		switch err {
+		case context.Canceled:
+			return errwrap.Wrapf(err, "ctx canceled")
+		case context.DeadlineExceeded:
+			return errwrap.Wrapf(err, "ctx deadline exceeded")
+		case rpctypes.ErrEmptyKey:
+			return errwrap.Wrapf(err, "client-side error")
+		default:
+			return errwrap.Wrapf(err, "invalid endpoints")
+		}
+	}
+	return nil
+}
+
+// txn runs a txn in etcd.
+func (obj *Fs) txn(ifcmps []etcd.Cmp, thenops, elseops []etcd.Op) (*etcd.TxnResponse, error) {
+	ctx, cancel := context.WithTimeout(context.Background(), EtcdTimeout)
+	resp, err := obj.Client.Txn(ctx).If(ifcmps...).Then(thenops...).Else(elseops...).Commit()
+	cancel()
+	return resp, err
+}
+
+// hash is a small helper that does the hashing for us.
+func (obj *Fs) hash(input []byte) (string, error) {
+	var h hash.Hash
+	switch obj.Hash {
+	// TODO: add other hashes
+	case "sha256":
+		h = sha256.New()
+	default:
+		return "", fmt.Errorf("hash does not exist")
+	}
+	src := bytes.NewReader(input)
+	if _, err := io.Copy(h, src); err != nil {
+		return "", err
+	}
+	return hex.EncodeToString(h.Sum(nil)), nil
+}
+
+// sync overwrites the superblock with whatever version we have stored.
+func (obj *Fs) sync() error {
+	b := bytes.Buffer{}
+	e := gob.NewEncoder(&b)
+	err := e.Encode(&obj.sb) // pass with &
+	if err != nil {
+		return errwrap.Wrapf(err, "gob failed to encode")
+	}
+	//base64.StdEncoding.EncodeToString(b.Bytes())
+	return obj.put(obj.Metadata, b.Bytes())
+}
+
+// mount downloads the initial cache of metadata, including the *file tree.
+// Since there's no explicit mount API in the afero.Fs interface, we hide this
+// method inside any operation that might do any real work, and make it
+// idempotent so that it can be called as much as we want. If there's no
+// metadata found (superblock) then we create one.
+func (obj *Fs) mount() error {
+	if obj.mounted {
+		return nil
+	}
+
+	result, err := obj.get(obj.Metadata) // download the metadata...
+	if err != nil {
+		return err
+	}
+	if result == nil || len(result) == 0 { // nothing found, create the fs
+		if obj.Debug {
+			log.Printf("debug: mount: creating new fs at: %s", obj.Metadata)
+		}
+		// trim any trailing slashes from DataPrefix
+		for strings.HasSuffix(obj.DataPrefix, "/") {
+			obj.DataPrefix = strings.TrimSuffix(obj.DataPrefix, "/")
+		}
+		if obj.DataPrefix == "" {
+			obj.DataPrefix = DefaultDataPrefix
+		}
+		if obj.Hash == "" {
+			obj.Hash = DefaultHash
+		}
+		// test run an empty string to see if our hash selection works!
+		if _, err := obj.hash([]byte("")); err != nil {
+			return fmt.Errorf("cannot hash with %s", obj.Hash)
+		}
+
+		obj.sb = &superBlock{
+			DataPrefix: obj.DataPrefix,
+			Hash:       obj.Hash,
+			Tree: &File{ // include a root directory
+				fs:   obj,
+				Path: "", // root dir is "" (empty string)
+				Mode: os.ModeDir,
+			},
+		}
+		if err := obj.sync(); err != nil {
+			return err
+		}
+
+		obj.mounted = true
+		return nil
+	}
+
+	if obj.Debug {
+		log.Printf("debug: mount: opening old fs at: %s", obj.Metadata)
+	}
+	sb, exists := result[obj.Metadata]
+	if !exists {
+		return fmt.Errorf("could not find metadata") // programming error?
+	}
+
+	// decode into obj.sb
+	//bb, err := base64.StdEncoding.DecodeString(str)
+	//if err != nil {
+	//	return errwrap.Wrapf(err, "base64 failed to decode")
+	//}
+	//b := bytes.NewBuffer(bb)
+	b := bytes.NewBuffer(sb)
+	d := gob.NewDecoder(b)
+	if err := d.Decode(&obj.sb); err != nil { // pass with &
+		return errwrap.Wrapf(err, "gob failed to decode")
+	}
+
+	if obj.DataPrefix != "" && obj.DataPrefix != obj.sb.DataPrefix {
+		return fmt.Errorf("the DataPrefix mount option `%s` does not match the remote value of `%s`", obj.DataPrefix, obj.sb.DataPrefix)
+	}
+	if obj.Hash != "" && obj.Hash != obj.sb.Hash {
+		return fmt.Errorf("the Hash mount option `%s` does not match the remote value of `%s`", obj.Hash, obj.sb.Hash)
+	}
+	// if all checks passed, copy these values down locally
+	obj.DataPrefix = obj.sb.DataPrefix
+	obj.Hash = obj.sb.Hash
+
+	// hook up file system pointers to each element in the tree structure
+	obj.traverse(obj.sb.Tree)
+
+	obj.mounted = true
+	return nil
+}
+
+// traverse adds the file system pointer to each element in the tree structure.
+func (obj *Fs) traverse(node *File) {
+	if node == nil {
+		return
+	}
+	node.fs = obj
+	for _, n := range node.Children {
+		obj.traverse(n)
+	}
+}
+
+// find returns the file node corresponding to this absolute path if it exists.
+func (obj *Fs) find(absPath string) (*File, error) { // TODO: function naming?
+	if absPath == "" {
+		return nil, fmt.Errorf("empty path specified")
+	}
+	if !strings.HasPrefix(absPath, "/") {
+		return nil, fmt.Errorf("invalid input path (not absolute)")
+	}
+
+	node := obj.sb.Tree
+	if node == nil {
+		return nil, ErrNotExist // no nodes exist yet, not even root dir
+	}
+
+	var x string // first value
+	sp := PathSplit(absPath)
+	if x, sp = sp[0], sp[1:]; x != node.Path {
+		return nil, fmt.Errorf("root values do not match") // TODO: panic?
+	}
+
+	for _, p := range sp {
+		n, exists := node.findNode(p)
+		if !exists {
+			return nil, ErrNotExist
+		}
+		node = n // descend into this node
+	}
+
+	return node, nil
+}
+
+// Name returns the name of this filesystem.
+func (obj *Fs) Name() string { return "etcdfs" }
+
+// URI returns a URI representing this particular filesystem.
+func (obj *Fs) URI() string {
+	return fmt.Sprintf("%s://%s", obj.Name(), obj.Metadata)
+}
+
+// Create creates a new file.
+func (obj *Fs) Create(name string) (afero.File, error) {
+	if err := obj.mount(); err != nil {
+		return nil, err
+	}
+	return fileCreate(obj, name)
+}
+
+// Mkdir makes a new directory.
+func (obj *Fs) Mkdir(name string, perm os.FileMode) error {
+	if err := obj.mount(); err != nil {
+		return err
+	}
+	if name == "" {
+		return fmt.Errorf("invalid input path")
+	}
+	if !strings.HasPrefix(name, "/") {
+		return fmt.Errorf("invalid input path (not absolute)")
+	}
+
+	// remove possible trailing slashes
+	cleanPath := path.Clean(name)
+
+	for strings.HasSuffix(cleanPath, "/") { // bonus clean for "/" as input
+		cleanPath = strings.TrimSuffix(cleanPath, "/")
+	}
+
+	if cleanPath == "" {
+		if obj.sb.Tree == nil {
+			return fmt.Errorf("woops, missing root directory")
+		}
+		return ErrExist // root directory already exists
+	}
+
+	// try to add node to tree by first finding the parent node
+	parentPath, dirPath := path.Split(cleanPath) // looking for this
+
+	f := &File{
+		fs:   obj,
+		Path: dirPath,
+		Mode: os.ModeDir,
+		// TODO: add perm to struct or let chmod below do it
+	}
+
+	node, err := obj.find(parentPath)
+	if err != nil { // might be ErrNotExist
+		return err
+	}
+
+	fi, err := node.Stat()
+	if err != nil {
+		return err
+	}
+	if !fi.IsDir() { // is the parent a suitable home?
+		return &os.PathError{Op: "mkdir", Path: name, Err: syscall.ENOTDIR}
+	}
+
+	_, exists := node.findNode(dirPath) // does file already exist inside?
+	if exists {
+		return ErrExist
+	}
+
+	// add to parent
+	node.Children = append(node.Children, f)
+
+	// push new file up if not on server, and then push up the metadata
+	if err := f.Sync(); err != nil {
+		return err
+	}
+
+	return obj.Chmod(name, perm)
+}
+
+// MkdirAll creates a directory named path, along with any necessary parents,
+// and returns nil, or else returns an error. The permission bits perm are used
+// for all directories that MkdirAll creates. If path is already a directory,
+// MkdirAll does nothing and returns nil.
+func (obj *Fs) MkdirAll(path string, perm os.FileMode) error {
+	if err := obj.mount(); err != nil {
+		return err
+	}
+	// Copied mostly verbatim from golang stdlib.
+	// Fast path: if we can tell whether path is a directory or file, stop
+	// with success or error.
+	dir, err := obj.Stat(path)
+	if err == nil {
+		if dir.IsDir() {
+			return nil
+		}
+		return &os.PathError{Op: "mkdir", Path: path, Err: syscall.ENOTDIR}
+	}
+
+	// Slow path: make sure parent exists and then call Mkdir for path.
+	i := len(path)
+	for i > 0 && IsPathSeparator(path[i-1]) { // Skip trailing path separator.
+		i--
+	}
+
+	j := i
+	for j > 0 && !IsPathSeparator(path[j-1]) { // Scan backward over element.
+		j--
+	}
+
+	if j > 1 {
+		// Create parent
+		err = obj.MkdirAll(path[0:j-1], perm)
+		if err != nil {
+			return err
+		}
+	}
+
+	// Parent now exists; invoke Mkdir and use its result.
+	err = obj.Mkdir(path, perm)
+	if err != nil {
+		// Handle arguments like "foo/." by
+		// double-checking that directory doesn't exist.
+		dir, err1 := obj.Lstat(path)
+		if err1 == nil && dir.IsDir() {
+			return nil
+		}
+		return err
+	}
+	return nil
+}
+
+// Open opens a path. It will be opened read-only.
+func (obj *Fs) Open(name string) (afero.File, error) {
+	if err := obj.mount(); err != nil {
+		return nil, err
+	}
+	return fileOpen(obj, name) // this opens as read-only
+}
+
+// OpenFile opens a path with a particular flag and permission.
+func (obj *Fs) OpenFile(name string, flag int, perm os.FileMode) (afero.File, error) {
+	if err := obj.mount(); err != nil {
+		return nil, err
+	}
+
+	chmod := false
+	f, err := fileOpen(obj, name)
+	if os.IsNotExist(err) && (flag&os.O_CREATE > 0) {
+		f, err = fileCreate(obj, name)
+		chmod = true
+	}
+	if err != nil {
+		return nil, err
+	}
+	f.readOnly = (flag == os.O_RDONLY)
+
+	if flag&os.O_APPEND > 0 {
+		if _, err := f.Seek(0, os.SEEK_END); err != nil {
+			f.Close()
+			return nil, err
+		}
+	}
+	if flag&os.O_TRUNC > 0 && flag&(os.O_RDWR|os.O_WRONLY) > 0 {
+		if err := f.Truncate(0); err != nil {
+			f.Close()
+			return nil, err
+		}
+	}
+	if chmod {
+		// TODO: the golang stdlib doesn't check this error, should we?
+		if err := obj.Chmod(name, perm); err != nil {
+			return f, err // TODO: should we return the file handle?
+		}
+	}
+	return f, nil
+}
+
+// Remove removes a path.
+func (obj *Fs) Remove(name string) error {
+	if err := obj.mount(); err != nil {
+		return err
+	}
+	if name == "" {
+		return fmt.Errorf("invalid input path")
+	}
+	if !strings.HasPrefix(name, "/") {
+		return fmt.Errorf("invalid input path (not absolute)")
+	}
+
+	// remove possible trailing slashes
+	cleanPath := path.Clean(name)
+
+	for strings.HasSuffix(cleanPath, "/") { // bonus clean for "/" as input
+		cleanPath = strings.TrimSuffix(cleanPath, "/")
+	}
+
+	if cleanPath == "" {
+		return fmt.Errorf("can't remove root")
+	}
+
+	f, err := obj.find(name) // get the file
+	if err != nil {
+		return err
+	}
+
+	if len(f.Children) > 0 { // this file or dir has children, can't remove!
+		return &os.PathError{Op: "remove", Path: name, Err: syscall.ENOTEMPTY}
+	}
+
+	// find the parent node
+	parentPath, filePath := path.Split(cleanPath) // looking for this
+
+	node, err := obj.find(parentPath)
+	if err != nil { // might be ErrNotExist
+		if os.IsNotExist(err) { // race! must have just disappeared
+			return nil
+		}
+		return err
+	}
+
+	var index = -1 // int
+	for i, n := range node.Children {
+		if n.Path == filePath {
+			index = i // found here!
+			break
+		}
+	}
+	if index == -1 {
+		return fmt.Errorf("programming error")
+	}
+	// remove from list
+	node.Children = append(node.Children[:index], node.Children[index+1:]...)
+	return obj.sync()
+}
+
+// RemoveAll removes path and any children it contains. It removes everything it
+// can but returns the first error it encounters. If the path does not exist,
+// RemoveAll returns nil (no error).
+func (obj *Fs) RemoveAll(path string) error {
+	if err := obj.mount(); err != nil {
+		return err
+	}
+
+	// Simple case: if Remove works, we're done.
+	err := obj.Remove(path)
+	if err == nil || os.IsNotExist(err) {
+		return nil
+	}
+
+	// Otherwise, is this a directory we need to recurse into?
+	dir, serr := obj.Lstat(path)
+	if serr != nil {
+		// TODO: I didn't check this logic thoroughly (edge cases?)
+		if serr, ok := serr.(*os.PathError); ok && (os.IsNotExist(serr.Err) || serr.Err == syscall.ENOTDIR) {
+			return nil
+		}
+		return serr
+	}
+	if !dir.IsDir() {
+		// Not a directory; return the error from Remove.
+		return err
+	}
+
+	// Directory.
+	fd, err := obj.Open(path)
+	if err != nil {
+		if os.IsNotExist(err) {
+			// Race. It was deleted between the Lstat and Open.
+			// Return nil per RemoveAll's docs.
+			return nil
+		}
+		return err
+	}
+
+	// Remove contents & return first error.
+	err = nil
+	for {
+		// TODO: why not do this in one shot? is there a syscall limit?
+		names, err1 := fd.Readdirnames(100)
+		for _, name := range names {
+			err1 := obj.RemoveAll(path + string(PathSeparator) + name)
+			if err == nil {
+				err = err1
+			}
+		}
+		if err1 == io.EOF {
+			break
+		}
+		// If Readdirnames returned an error, use it.
+		if err == nil {
+			err = err1
+		}
+		if len(names) == 0 {
+			break
+		}
+	}
+
+	// Close directory, because windows won't remove opened directory.
+	fd.Close()
+
+	// Remove directory.
+	err1 := obj.Remove(path)
+	if err1 == nil || os.IsNotExist(err1) {
+		return nil
+	}
+	if err == nil {
+		err = err1
+	}
+	return err
+}
+
+// Rename moves or renames a file or directory.
+// TODO: seems it's okay to move files or directories, but you can't clobber dirs
+// but you can clobber single files. a dir can't clobber a file and a file can't
+// clobber a dir. but a file can clobber another file but a dir can't clobber
+// another dir. you can also transplant dirs or files into other dirs.
+func (obj *Fs) Rename(oldname, newname string) error {
+	// XXX: do we need to check if dest path is inside src path?
+	// XXX: if dirs/files are next to each other, do we mess up the .Children list of the common parent?
+	if err := obj.mount(); err != nil {
+		return err
+	}
+
+	if oldname == newname {
+		return nil
+	}
+	if oldname == "" || newname == "" {
+		return fmt.Errorf("invalid input path")
+	}
+	if !strings.HasPrefix(oldname, "/") || !strings.HasPrefix(newname, "/") {
+		return fmt.Errorf("invalid input path (not absolute)")
+	}
+
+	// remove possible trailing slashes
+	srcCleanPath := path.Clean(oldname)
+	dstCleanPath := path.Clean(newname)
+
+	src, err := obj.find(srcCleanPath) // get the file
+	if err != nil {
+		return err
+	}
+
+	srcInfo, err := src.Stat()
+	if err != nil {
+		return err
+	}
+
+	srcParentPath, srcName := path.Split(srcCleanPath) // looking for this
+	parent, err := obj.find(srcParentPath)
+	if err != nil { // might be ErrNotExist
+		return err
+	}
+	var rmi = -1 // index of node to remove from parent
+	// find the thing to be deleted
+	for i, n := range parent.Children {
+		if n.Path == srcName {
+			rmi = i // found here!
+			break
+		}
+	}
+	if rmi == -1 {
+		return fmt.Errorf("programming error")
+	}
+
+	dst, err := obj.find(dstCleanPath) // does the destination already exist?
+	if err != nil && !os.IsNotExist(err) {
+		return err
+	}
+	if err == nil { // dst exists!
+		dstInfo, err := dst.Stat()
+		if err != nil {
+			return err
+		}
+
+		// dir's can clobber anything or be clobbered apparently
+		if srcInfo.IsDir() || dstInfo.IsDir() {
+			return ErrExist // dir's can't clobber anything
+		}
+
+		// remove from list by index
+		parent.Children = append(parent.Children[:rmi], parent.Children[rmi+1:]...)
+
+		// we're a file clobbering another file...
+		// move file content from src -> dst and then delete src
+		// TODO: run a dst.Close() for extra safety first?
+		save := dst.Path // save the "name"
+		*dst = *src      // TODO: is this safe?
+		dst.Path = save  // "rename" it
+
+	} else { // dst does not exist
+
+		// check if the dst's parent exists and is a dir, if not, error
+		// if it is a dir, add src as a child to it and then delete src
+		dstParentPath, dstName := path.Split(dstCleanPath) // looking for this
+		node, err := obj.find(dstParentPath)
+		if err != nil { // might be ErrNotExist
+			return err
+		}
+		fi, err := node.Stat()
+		if err != nil {
+			return err
+		}
+		if !fi.IsDir() { // is the parent a suitable home?
+			return &os.LinkError{Op: "rename", Old: oldname, New: newname, Err: syscall.ENOTDIR}
+		}
+
+		// remove from list by index
+		parent.Children = append(parent.Children[:rmi], parent.Children[rmi+1:]...)
+
+		src.Path = dstName                         // "rename" it
+		node.Children = append(node.Children, src) // "copied"
+	}
+
+	return obj.sync() // push up metadata changes
+}
+
+// Stat returns some information about the particular path.
+func (obj *Fs) Stat(name string) (os.FileInfo, error) {
+	if err := obj.mount(); err != nil {
+		return nil, err
+	}
+	if !strings.HasPrefix(name, "/") {
+		return nil, fmt.Errorf("invalid input path (not absolute)")
+	}
+
+	f, err := obj.find(name) // get the file
+	if err != nil {
+		return nil, err
+	}
+	return f.Stat()
+}
+
+// Lstat does exactly the same as Stat because we currently do not support
+// symbolic links.
+func (obj *Fs) Lstat(name string) (os.FileInfo, error) {
+	if err := obj.mount(); err != nil {
+		return nil, err
+	}
+	// TODO: we don't have symbolic links in our fs, so we pass this to stat
+	return obj.Stat(name)
+}
+
+// Chmod changes the mode of a file.
+func (obj *Fs) Chmod(name string, mode os.FileMode) error {
+	if err := obj.mount(); err != nil {
+		return err
+	}
+	if !strings.HasPrefix(name, "/") {
+		return fmt.Errorf("invalid input path (not absolute)")
+	}
+
+	f, err := obj.find(name) // get the file
+	if err != nil {
+		return err
+	}
+
+	f.Mode = f.Mode | mode // XXX: what is the correct way to do this?
+	return f.Sync()        // push up the changed metadata
+}
+
+// Chtimes changes the access and modification times of the named file, similar
+// to the Unix utime() or utimes() functions. The underlying filesystem may
+// truncate or round the values to a less precise time unit. If there is an
+// error, it will be of type *PathError.
+// FIXME: make sure everything we error is a *PathError
+// TODO: atime is not currently implement and so it is silently ignored.
+func (obj *Fs) Chtimes(name string, atime time.Time, mtime time.Time) error {
+	if err := obj.mount(); err != nil {
+		return err
+	}
+	if !strings.HasPrefix(name, "/") {
+		return fmt.Errorf("invalid input path (not absolute)")
+	}
+
+	f, err := obj.find(name) // get the file
+	if err != nil {
+		return err
+	}
+
+	f.ModTime = mtime
+	// TODO: add atime
+	return f.Sync() // push up the changed metadata
+}
+
+// PathSplit splits a path into an array of tokens excluding any trailing empty
+// tokens.
+func PathSplit(p string) []string {
+	if p == "/" { // TODO: can't this all be expressed nicely in one line?
+		return []string{""}
+	}
+	return strings.Split(path.Clean(p), "/")
+}

etcd/fs/fs_test.go

@@ -0,0 +1,227 @@
+// Mgmt
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
+// Written by James Shubin <james@shubin.ca> and the project contributors
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+package fs_test // named this way to make it easier for examples
+
+import (
+	"io"
+	"testing"
+
+	"github.com/purpleidea/mgmt/etcd"
+	etcdfs "github.com/purpleidea/mgmt/etcd/fs"
+	"github.com/purpleidea/mgmt/util"
+
+	"github.com/spf13/afero"
+)
+
+// XXX: spawn etcd for this test, like `cdtmpmkdir && etcd` and then kill it...
+// XXX: write a bunch more tests to test this
+
+// TODO: apparently using 0666 is equivalent to respecting the current umask
+const (
+	umask      = 0666
+	superblock = "/some/superblock" // TODO: generate randomly per test?
+)
+
+func TestFs1(t *testing.T) {
+	etcdClient := &etcd.ClientEtcd{
+		Seeds: []string{"localhost:2379"}, // endpoints
+	}
+
+	if err := etcdClient.Connect(); err != nil {
+		t.Logf("client connection error: %+v", err)
+		return
+	}
+	defer etcdClient.Destroy()
+
+	etcdFs := &etcdfs.Fs{
+		Client:     etcdClient.GetClient(),
+		Metadata:   superblock,
+		DataPrefix: etcdfs.DefaultDataPrefix,
+	}
+	//var etcdFs afero.Fs = NewEtcdFs()
+
+	if err := etcdFs.Mkdir("/", umask); err != nil {
+		t.Logf("error: %+v", err)
+		if err != etcdfs.ErrExist {
+			return
+		}
+	}
+
+	if err := etcdFs.Mkdir("/tmp", umask); err != nil {
+		t.Logf("error: %+v", err)
+		if err != etcdfs.ErrExist {
+			return
+		}
+	}
+
+	fi, err := etcdFs.Stat("/tmp")
+	if err != nil {
+		t.Logf("stat error: %+v", err)
+		return
+	}
+
+	t.Logf("fi: %+v", fi)
+	t.Logf("isdir: %t", fi.IsDir())
+
+	f, err := etcdFs.Create("/tmp/foo")
+	if err != nil {
+		t.Logf("error: %+v", err)
+		return
+	}
+
+	t.Logf("handle: %+v", f)
+
+	i, err := f.WriteString("hello world!\n")
+	if err != nil {
+		t.Logf("error: %+v", err)
+		return
+	}
+	t.Logf("wrote: %d", i)
+
+	if err := etcdFs.Mkdir("/tmp/d1", umask); err != nil {
+		t.Logf("error: %+v", err)
+		if err != etcdfs.ErrExist {
+			return
+		}
+	}
+
+	if err := etcdFs.Rename("/tmp/foo", "/tmp/bar"); err != nil {
+		t.Logf("rename error: %+v", err)
+		return
+	}
+
+	//f2, err := etcdFs.Create("/tmp/bar")
+	//if err != nil {
+	//	t.Logf("error: %+v", err)
+	//	return
+	//}
+
+	//i2, err := f2.WriteString("hello bar!\n")
+	//if err != nil {
+	//	t.Logf("error: %+v", err)
+	//	return
+	//}
+	//t.Logf("wrote: %d", i2)
+
+	dir, err := etcdFs.Open("/tmp")
+	if err != nil {
+		t.Logf("error: %+v", err)
+		return
+	}
+	names, err := dir.Readdirnames(-1)
+	if err != nil && err != io.EOF {
+		t.Logf("error: %+v", err)
+		return
+	}
+	for _, name := range names {
+		t.Logf("name in /tmp: %+v", name)
+	}
+
+	//dir, err := etcdFs.Open("/")
+	//if err != nil {
+	//	t.Logf("error: %+v", err)
+	//	return
+	//}
+	//names, err := dir.Readdirnames(-1)
+	//if err != nil && err != io.EOF {
+	//	t.Logf("error: %+v", err)
+	//	return
+	//}
+	//for _, name := range names {
+	//	t.Logf("name in /: %+v", name)
+	//}
+}
+
+func TestFs2(t *testing.T) {
+	etcdClient := &etcd.ClientEtcd{
+		Seeds: []string{"localhost:2379"}, // endpoints
+	}
+
+	if err := etcdClient.Connect(); err != nil {
+		t.Logf("client connection error: %+v", err)
+		return
+	}
+	defer etcdClient.Destroy()
+
+	etcdFs := &etcdfs.Fs{
+		Client:     etcdClient.GetClient(),
+		Metadata:   superblock,
+		DataPrefix: etcdfs.DefaultDataPrefix,
+	}
+
+	tree, err := util.FsTree(etcdFs, "/")
+	if err != nil {
+		t.Errorf("tree error: %+v", err)
+		return
+	}
+	t.Logf("tree: \n%s", tree)
+
+	tree2, err := util.FsTree(etcdFs, "/tmp")
+	if err != nil {
+		t.Errorf("tree2 error: %+v", err)
+		return
+	}
+	t.Logf("tree2: \n%s", tree2)
+}
+
+func TestFs3(t *testing.T) {
+	etcdClient := &etcd.ClientEtcd{
+		Seeds: []string{"localhost:2379"}, // endpoints
+	}
+
+	if err := etcdClient.Connect(); err != nil {
+		t.Logf("client connection error: %+v", err)
+		return
+	}
+	defer etcdClient.Destroy()
+
+	etcdFs := &etcdfs.Fs{
+		Client:     etcdClient.GetClient(),
+		Metadata:   superblock,
+		DataPrefix: etcdfs.DefaultDataPrefix,
+	}
+
+	tree, err := util.FsTree(etcdFs, "/")
+	if err != nil {
+		t.Errorf("tree error: %+v", err)
+		return
+	}
+	t.Logf("tree: \n%s", tree)
+
+	var memFs = afero.NewMemMapFs()
+
+	if err := util.CopyFs(etcdFs, memFs, "/", "/", false); err != nil {
+		t.Errorf("CopyFs error: %+v", err)
+		return
+	}
+	if err := util.CopyFs(etcdFs, memFs, "/", "/", true); err != nil {
+		t.Errorf("CopyFs2 error: %+v", err)
+		return
+	}
+	if err := util.CopyFs(etcdFs, memFs, "/", "/tmp/d1/", false); err != nil {
+		t.Errorf("CopyFs3 error: %+v", err)
+		return
+	}
+
+	tree2, err := util.FsTree(memFs, "/")
+	if err != nil {
+		t.Errorf("tree2 error: %+v", err)
+		return
+	}
+	t.Logf("tree2: \n%s", tree2)
+}

etcd/fs/util.go

@@ -0,0 +1,88 @@
+// Mgmt
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
+// Written by James Shubin <james@shubin.ca> and the project contributors
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+package fs
+
+import (
+	"os"
+	"path/filepath"
+
+	"github.com/spf13/afero"
+)
+
+// ReadAll reads from r until an error or EOF and returns the data it read.
+// A successful call returns err == nil, not err == EOF. Because ReadAll is
+// defined to read from src until EOF, it does not treat an EOF from Read
+// as an error to be reported.
+//func ReadAll(r io.Reader) ([]byte, error) {
+//	return afero.ReadAll(r)
+//}
+
+// ReadDir reads the directory named by dirname and returns
+// a list of sorted directory entries.
+func (obj *Fs) ReadDir(dirname string) ([]os.FileInfo, error) {
+	return afero.ReadDir(obj, dirname)
+}
+
+// ReadFile reads the file named by filename and returns the contents.
+// A successful call returns err == nil, not err == EOF. Because ReadFile
+// reads the whole file, it does not treat an EOF from Read as an error
+// to be reported.
+func (obj *Fs) ReadFile(filename string) ([]byte, error) {
+	return afero.ReadFile(obj, filename)
+}
+
+// TempDir creates a new temporary directory in the directory dir
+// with a name beginning with prefix and returns the path of the
+// new directory.  If dir is the empty string, TempDir uses the
+// default directory for temporary files (see os.TempDir).
+// Multiple programs calling TempDir simultaneously
+// will not choose the same directory.  It is the caller's responsibility
+// to remove the directory when no longer needed.
+func (obj *Fs) TempDir(dir, prefix string) (name string, err error) {
+	return afero.TempDir(obj, dir, prefix)
+}
+
+// TempFile creates a new temporary file in the directory dir
+// with a name beginning with prefix, opens the file for reading
+// and writing, and returns the resulting *File.
+// If dir is the empty string, TempFile uses the default directory
+// for temporary files (see os.TempDir).
+// Multiple programs calling TempFile simultaneously
+// will not choose the same file.  The caller can use f.Name()
+// to find the pathname of the file.  It is the caller's responsibility
+// to remove the file when no longer needed.
+func (obj *Fs) TempFile(dir, prefix string) (f afero.File, err error) {
+	return afero.TempFile(obj, dir, prefix)
+}
+
+// WriteFile writes data to a file named by filename.
+// If the file does not exist, WriteFile creates it with permissions perm;
+// otherwise WriteFile truncates it before writing.
+func (obj *Fs) WriteFile(filename string, data []byte, perm os.FileMode) error {
+	return afero.WriteFile(obj, filename, data, perm)
+}
+
+// Walk walks the file tree rooted at root, calling walkFn for each file or
+// directory in the tree, including root. All errors that arise visiting files
+// and directories are filtered by walkFn. The files are walked in lexical
+// order, which makes the output deterministic but means that for very
+// large directories Walk can be inefficient.
+// Walk does not follow symbolic links.
+func (obj *Fs) Walk(root string, walkFn filepath.WalkFunc) error {
+	return afero.Walk(obj, root, walkFn)
+}

etcd/interfaces.go

@@ -0,0 +1,30 @@
+// Mgmt
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
+// Written by James Shubin <james@shubin.ca> and the project contributors
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+package etcd
+
+import (
+	etcd "github.com/coreos/etcd/clientv3" // "clientv3"
+)
+
+// Client provides a simple interface specification for client requests. Both
+// EmbdEtcd and ClientEtcd implement this.
+type Client interface {
+	// TODO: add more method signatures
+	Get(path string, opts ...etcd.OpOption) (map[string]string, error)
+	Txn(ifcmps []etcd.Cmp, thenops, elseops []etcd.Op) (*etcd.TxnResponse, error)
+}

etcd/methods.go

@@ -1,5 +1,5 @@
 // Mgmt
-// Copyright (C) 2013-2017+ James Shubin and the project contributors
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
 // Written by James Shubin <james@shubin.ca> and the project contributors
 //
 // This program is free software: you can redistribute it and/or modify
@@ -39,7 +39,7 @@ func Nominate(obj *EmbdEtcd, hostname string, urls etcdtypes.URLs) error {
 		defer log.Printf("Trace: Etcd: Nominate(%v): Finished!", hostname)
 	}
 	// nominate someone to be a server
-	nominate := fmt.Sprintf("/%s/nominated/%s", NS, hostname)
+	nominate := fmt.Sprintf("%s/nominated/%s", NS, hostname)
 	ops := []etcd.Op{} // list of ops in this txn
 	if urls != nil {
 		ops = append(ops, etcd.OpPut(nominate, urls.String())) // TODO: add a TTL? (etcd.WithLease)
@@ -57,7 +57,7 @@ func Nominate(obj *EmbdEtcd, hostname string, urls etcdtypes.URLs) error {
 // Nominated returns a urls map of nominated etcd server volunteers.
 // NOTE: I know 'nominees' might be more correct, but is less consistent here
 func Nominated(obj *EmbdEtcd) (etcdtypes.URLsMap, error) {
-	path := fmt.Sprintf("/%s/nominated/", NS)
+	path := fmt.Sprintf("%s/nominated/", NS)
 	keyMap, err := obj.Get(path, etcd.WithPrefix()) // map[string]string, bool
 	if err != nil {
 		return nil, fmt.Errorf("nominated isn't available: %v", err)
@@ -90,7 +90,7 @@ func Volunteer(obj *EmbdEtcd, urls etcdtypes.URLs) error {
 		defer log.Printf("Trace: Etcd: Volunteer(%v): Finished!", obj.hostname)
 	}
 	// volunteer to be a server
-	volunteer := fmt.Sprintf("/%s/volunteers/%s", NS, obj.hostname)
+	volunteer := fmt.Sprintf("%s/volunteers/%s", NS, obj.hostname)
 	ops := []etcd.Op{} // list of ops in this txn
 	if urls != nil {
 		// XXX: adding a TTL is crucial! (i think)
@@ -112,7 +112,7 @@ func Volunteers(obj *EmbdEtcd) (etcdtypes.URLsMap, error) {
 		log.Printf("Trace: Etcd: Volunteers()")
 		defer log.Printf("Trace: Etcd: Volunteers(): Finished!")
 	}
-	path := fmt.Sprintf("/%s/volunteers/", NS)
+	path := fmt.Sprintf("%s/volunteers/", NS)
 	keyMap, err := obj.Get(path, etcd.WithPrefix())
 	if err != nil {
 		return nil, fmt.Errorf("volunteers aren't available: %v", err)
@@ -145,7 +145,7 @@ func AdvertiseEndpoints(obj *EmbdEtcd, urls etcdtypes.URLs) error {
 		defer log.Printf("Trace: Etcd: AdvertiseEndpoints(%v): Finished!", obj.hostname)
 	}
 	// advertise endpoints
-	endpoints := fmt.Sprintf("/%s/endpoints/%s", NS, obj.hostname)
+	endpoints := fmt.Sprintf("%s/endpoints/%s", NS, obj.hostname)
 	ops := []etcd.Op{} // list of ops in this txn
 	if urls != nil {
 		// TODO: add a TTL? (etcd.WithLease)
@@ -167,7 +167,7 @@ func Endpoints(obj *EmbdEtcd) (etcdtypes.URLsMap, error) {
 		log.Printf("Trace: Etcd: Endpoints()")
 		defer log.Printf("Trace: Etcd: Endpoints(): Finished!")
 	}
-	path := fmt.Sprintf("/%s/endpoints/", NS)
+	path := fmt.Sprintf("%s/endpoints/", NS)
 	keyMap, err := obj.Get(path, etcd.WithPrefix())
 	if err != nil {
 		return nil, fmt.Errorf("endpoints aren't available: %v", err)
@@ -199,7 +199,7 @@ func SetHostnameConverged(obj *EmbdEtcd, hostname string, isConverged bool) erro
 		log.Printf("Trace: Etcd: SetHostnameConverged(%s): %v", hostname, isConverged)
 		defer log.Printf("Trace: Etcd: SetHostnameConverged(%v): Finished!", hostname)
 	}
-	converged := fmt.Sprintf("/%s/converged/%s", NS, hostname)
+	converged := fmt.Sprintf("%s/converged/%s", NS, hostname)
 	op := []etcd.Op{etcd.OpPut(converged, fmt.Sprintf("%t", isConverged))}
 	if _, err := obj.Txn(nil, op, nil); err != nil { // TODO: do we need a skipConv flag here too?
 		return fmt.Errorf("set converged failed") // exit in progress?
@@ -213,7 +213,7 @@ func HostnameConverged(obj *EmbdEtcd) (map[string]bool, error) {
 		log.Printf("Trace: Etcd: HostnameConverged()")
 		defer log.Printf("Trace: Etcd: HostnameConverged(): Finished!")
 	}
-	path := fmt.Sprintf("/%s/converged/", NS)
+	path := fmt.Sprintf("%s/converged/", NS)
 	keyMap, err := obj.ComplexGet(path, true, etcd.WithPrefix()) // don't un-converge
 	if err != nil {
 		return nil, fmt.Errorf("converged values aren't available: %v", err)
@@ -239,7 +239,7 @@ func HostnameConverged(obj *EmbdEtcd) (map[string]bool, error) {
 // AddHostnameConvergedWatcher adds a watcher with a callback that runs on
 // hostname state changes.
 func AddHostnameConvergedWatcher(obj *EmbdEtcd, callbackFn func(map[string]bool) error) (func(), error) {
-	path := fmt.Sprintf("/%s/converged/", NS)
+	path := fmt.Sprintf("%s/converged/", NS)
 	internalCbFn := func(re *RE) error {
 		// TODO: get the value from the response, and apply delta...
 		// for now, just run a get operation which is easier to code!
@@ -258,7 +258,7 @@ func SetClusterSize(obj *EmbdEtcd, value uint16) error {
 		log.Printf("Trace: Etcd: SetClusterSize(): %v", value)
 		defer log.Printf("Trace: Etcd: SetClusterSize(): Finished!")
 	}
-	key := fmt.Sprintf("/%s/idealClusterSize", NS)
+	key := fmt.Sprintf("%s/idealClusterSize", NS)
 
 	if err := obj.Set(key, strconv.FormatUint(uint64(value), 10)); err != nil {
 		return fmt.Errorf("function SetClusterSize failed: %v", err) // exit in progress?
@@ -268,7 +268,7 @@ func SetClusterSize(obj *EmbdEtcd, value uint16) error {
 
 // GetClusterSize gets the ideal target cluster size of etcd peers.
 func GetClusterSize(obj *EmbdEtcd) (uint16, error) {
-	key := fmt.Sprintf("/%s/idealClusterSize", NS)
+	key := fmt.Sprintf("%s/idealClusterSize", NS)
 	keyMap, err := obj.Get(key)
 	if err != nil {
 		return 0, fmt.Errorf("function GetClusterSize failed: %v", err)
@@ -371,9 +371,8 @@ func Members(obj *EmbdEtcd) (map[uint64]string, error) {
 // Leader returns the current leader of the etcd server cluster.
 func Leader(obj *EmbdEtcd) (string, error) {
 	//obj.Connect(false) // TODO: ?
-	var err error
-	membersMap := make(map[uint64]string)
-	if membersMap, err = Members(obj); err != nil {
+	membersMap, err := Members(obj)
+	if err != nil {
 		return "", err
 	}
 	addresses := obj.LocalhostClientURLs() // heuristic, but probably correct

etcd/resources.go

@@ -1,5 +1,5 @@
 // Mgmt
-// Copyright (C) 2013-2017+ James Shubin and the project contributors
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
 // Written by James Shubin <james@shubin.ca> and the project contributors
 //
 // This program is free software: you can redistribute it and/or modify
@@ -34,7 +34,7 @@ import (
 // collection prefixes and filters that we care about...
 func WatchResources(obj *EmbdEtcd) chan error {
 	ch := make(chan error, 1) // buffer it so we can measure it
-	path := fmt.Sprintf("/%s/exported/", NS)
+	path := fmt.Sprintf("%s/exported/", NS)
 	callback := func(re *RE) error {
 		// TODO: is this even needed? it used to happen on conn errors
 		log.Printf("Etcd: Watch: Path: %v", path) // event
@@ -61,7 +61,7 @@ func WatchResources(obj *EmbdEtcd) chan error {
 
 // SetResources exports all of the resources which we pass in to etcd.
 func SetResources(obj *EmbdEtcd, hostname string, resourceList []resources.Res) error {
-	// key structure is /$NS/exported/$hostname/resources/$uid = $data
+	// key structure is $NS/exported/$hostname/resources/$uid = $data
 
 	var kindFilter []string // empty to get from everyone
 	hostnameFilter := []string{hostname}
@@ -83,7 +83,7 @@ func SetResources(obj *EmbdEtcd, hostname string, resourceList []resources.Res)
 			log.Fatalf("Etcd: SetResources: Error: Empty kind: %v", res.GetName())
 		}
 		uid := fmt.Sprintf("%s/%s", res.GetKind(), res.GetName())
-		path := fmt.Sprintf("/%s/exported/%s/resources/%s", NS, hostname, uid)
+		path := fmt.Sprintf("%s/exported/%s/resources/%s", NS, hostname, uid)
 		if data, err := resources.ResToB64(res); err == nil {
 			ifs = append(ifs, etcd.Compare(etcd.Value(path), "=", data)) // desired state
 			ops = append(ops, etcd.OpPut(path, data))
@@ -108,7 +108,7 @@ func SetResources(obj *EmbdEtcd, hostname string, resourceList []resources.Res)
 			log.Fatalf("Etcd: SetResources: Error: Empty kind: %v", res.GetName())
 		}
 		uid := fmt.Sprintf("%s/%s", res.GetKind(), res.GetName())
-		path := fmt.Sprintf("/%s/exported/%s/resources/%s", NS, hostname, uid)
+		path := fmt.Sprintf("%s/exported/%s/resources/%s", NS, hostname, uid)
 
 		if match(res, resourceList) { // if we match, no need to delete!
 			continue
@@ -134,10 +134,10 @@ func SetResources(obj *EmbdEtcd, hostname string, resourceList []resources.Res)
 // If the kindfilter or hostnameFilter is empty, then it assumes no filtering...
 // TODO: Expand this with a more powerful filter based on what we eventually
 // support in our collect DSL. Ideally a server side filter like WithFilter()
-// We could do this if the pattern was /$NS/exported/$kind/$hostname/$uid = $data.
+// We could do this if the pattern was $NS/exported/$kind/$hostname/$uid = $data.
 func GetResources(obj *EmbdEtcd, hostnameFilter, kindFilter []string) ([]resources.Res, error) {
-	// key structure is /$NS/exported/$hostname/resources/$uid = $data
-	path := fmt.Sprintf("/%s/exported/", NS)
+	// key structure is $NS/exported/$hostname/resources/$uid = $data
+	path := fmt.Sprintf("%s/exported/", NS)
 	resourceList := []resources.Res{}
 	keyMap, err := obj.Get(path, etcd.WithPrefix(), etcd.WithSort(etcd.SortByKey, etcd.SortAscend))
 	if err != nil {

etcd/scheduler/alphastrategy.go

@@ -0,0 +1,49 @@
+// Mgmt
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
+// Written by James Shubin <james@shubin.ca> and the project contributors
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+package scheduler // TODO: i'd like this to be a separate package, but cycles!
+
+import (
+	"fmt"
+	"sort"
+)
+
+func init() {
+	Register("alpha", func() Strategy { return &alphaStrategy{} }) // must register the func and name
+}
+
+type alphaStrategy struct {
+	// no state to store
+}
+
+// Schedule returns the first host out of a sorted group of available hostnames.
+func (obj *alphaStrategy) Schedule(hostnames map[string]string, opts *schedulerOptions) ([]string, error) {
+	if len(hostnames) <= 0 {
+		return nil, fmt.Errorf("strategy: cannot schedule from zero hosts")
+	}
+	if opts.maxCount <= 0 {
+		return nil, fmt.Errorf("strategy: cannot schedule with a max of zero")
+	}
+
+	sortedHosts := []string{}
+	for key := range hostnames {
+		sortedHosts = append(sortedHosts, key)
+	}
+	sort.Strings(sortedHosts)
+
+	return []string{sortedHosts[0]}, nil // pick first host
+}

etcd/scheduler/options.go

@@ -0,0 +1,100 @@
+// Mgmt
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
+// Written by James Shubin <james@shubin.ca> and the project contributors
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+package scheduler
+
+import (
+	"fmt"
+)
+
+// Option is a type that can be used to configure the scheduler.
+type Option func(*schedulerOptions)
+
+// schedulerOptions represents the different possible configurable options. Not
+// all options necessarily work for each scheduler strategy algorithm.
+type schedulerOptions struct {
+	debug       bool
+	logf        func(format string, v ...interface{})
+	strategy    Strategy
+	maxCount    int // TODO: should this be *int to know when it's set?
+	reuseLease  bool
+	sessionTTL  int // TODO: should this be *int to know when it's set?
+	hostsFilter []string
+	// TODO: add more options
+}
+
+// Debug specifies whether we should run in debug mode or not.
+func Debug(debug bool) Option {
+	return func(so *schedulerOptions) {
+		so.debug = debug
+	}
+}
+
+// Logf passes a logger function that we can use if so desired.
+func Logf(logf func(format string, v ...interface{})) Option {
+	return func(so *schedulerOptions) {
+		so.logf = logf
+	}
+}
+
+// StrategyKind sets the scheduler strategy used.
+func StrategyKind(strategy string) Option {
+	return func(so *schedulerOptions) {
+		f, exists := registeredStrategies[strategy]
+		if !exists {
+			panic(fmt.Sprintf("scheduler: undefined strategy: %s", strategy))
+		}
+		so.strategy = f()
+	}
+}
+
+// MaxCount is the maximum number of hosts that should get simultaneously
+// scheduled.
+func MaxCount(maxCount int) Option {
+	return func(so *schedulerOptions) {
+		if maxCount > 0 {
+			so.maxCount = maxCount
+		}
+	}
+}
+
+// ReuseLease specifies whether we should try and re-use the lease between runs.
+// Ordinarily it would get discarded with each new version (deploy) of the code.
+func ReuseLease(reuseLease bool) Option {
+	return func(so *schedulerOptions) {
+		so.reuseLease = reuseLease
+	}
+}
+
+// SessionTTL is the amount of time to delay before expiring a key on abrupt
+// host disconnect of if ReuseLease is true.
+func SessionTTL(sessionTTL int) Option {
+	return func(so *schedulerOptions) {
+		if sessionTTL > 0 {
+			so.sessionTTL = sessionTTL
+		}
+	}
+}
+
+// HostsFilter specifies a manual list of hosts, to use as a subset of whatever
+// was auto-discovered.
+// XXX: think more about this idea...
+func HostsFilter(hosts []string) Option {
+	return func(so *schedulerOptions) {
+		so.hostsFilter = hosts
+	}
+}

etcd/scheduler/rrstrategy.go

@@ -0,0 +1,84 @@
+// Mgmt
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
+// Written by James Shubin <james@shubin.ca> and the project contributors
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+package scheduler // TODO: i'd like this to be a separate package, but cycles!
+
+import (
+	"fmt"
+	"sort"
+
+	"github.com/purpleidea/mgmt/util"
+)
+
+func init() {
+	Register("rr", func() Strategy { return &rrStrategy{} }) // must register the func and name
+}
+
+type rrStrategy struct {
+	// some stored state
+	hosts []string
+}
+
+// Schedule returns hosts in round robin style from the available hostnames.
+func (obj *rrStrategy) Schedule(hostnames map[string]string, opts *schedulerOptions) ([]string, error) {
+	if len(hostnames) <= 0 {
+		return nil, fmt.Errorf("strategy: cannot schedule from zero hosts")
+	}
+	if opts.maxCount <= 0 {
+		return nil, fmt.Errorf("strategy: cannot schedule with a max of zero")
+	}
+
+	// always get a deterministic list of current hosts first...
+	sortedHosts := []string{}
+	for key := range hostnames {
+		sortedHosts = append(sortedHosts, key)
+	}
+	sort.Strings(sortedHosts)
+
+	if obj.hosts == nil {
+		obj.hosts = []string{} // initialize if needed
+	}
+
+	// add any new hosts we learned about, to the end of the list
+	for _, x := range sortedHosts {
+		if !util.StrInList(x, obj.hosts) {
+			obj.hosts = append(obj.hosts, x)
+		}
+	}
+
+	// remove any hosts we previouly knew about from the list
+	for ix := len(obj.hosts) - 1; ix >= 0; ix-- {
+		if !util.StrInList(obj.hosts[ix], sortedHosts) {
+			// delete entry at this index
+			obj.hosts = append(obj.hosts[:ix], obj.hosts[ix+1:]...)
+		}
+	}
+
+	// get the maximum number of hosts to return
+	max := len(obj.hosts)    // can't return more than we have
+	if opts.maxCount < max { // found a smaller limit
+		max = opts.maxCount
+	}
+
+	result := []string{}
+	// now return the number of needed hosts from the list
+	for i := 0; i < max; i++ {
+		result = append(result, obj.hosts[i])
+	}
+
+	return result, nil
+}

etcd/scheduler/scheduler.go

@@ -0,0 +1,564 @@
+// Mgmt
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
+// Written by James Shubin <james@shubin.ca> and the project contributors
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+// Package scheduler implements a distributed consensus scheduler with etcd.
+package scheduler
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"sort"
+	"strings"
+	"sync"
+
+	etcd "github.com/coreos/etcd/clientv3"
+	"github.com/coreos/etcd/clientv3/concurrency"
+	errwrap "github.com/pkg/errors"
+)
+
+const (
+	// DefaultSessionTTL is the number of seconds to wait before a dead or
+	// unresponsive host is removed from the scheduled pool.
+	DefaultSessionTTL = 10 // seconds
+
+	// DefaultMaxCount is the maximum number of hosts to schedule on if not
+	// specified.
+	DefaultMaxCount = 1 // TODO: what is the logical value to choose? +Inf?
+
+	hostnameJoinChar = "," // char used to join and split lists of hostnames
+)
+
+// ErrEndOfResults is a sentinel that represents no more results will be coming.
+var ErrEndOfResults = errors.New("scheduler: end of results")
+
+var schedulerLeases = make(map[string]etcd.LeaseID) // process lifetime in-memory lease store
+
+// schedulerResult represents output from the scheduler.
+type schedulerResult struct {
+	hosts []string
+	err   error
+}
+
+// Result is what is returned when you request a scheduler. You can call methods
+// on it, and it stores the necessary state while you're running. When one of
+// these is produced, the scheduler has already kicked off running for you
+// automatically.
+type Result struct {
+	results   chan *schedulerResult
+	closeFunc func() // run this when you're done with the scheduler // TODO: replace with an input `context`
+}
+
+// Next returns the next output from the scheduler when it changes. This blocks
+// until a new value is available, which is why you may wish to use a context to
+// cancel any read from this. It returns ErrEndOfResults if the scheduler shuts
+// down.
+func (obj *Result) Next(ctx context.Context) ([]string, error) {
+	select {
+	case val, ok := <-obj.results:
+		if !ok {
+			return nil, ErrEndOfResults
+		}
+		return val.hosts, val.err
+
+	case <-ctx.Done():
+		return nil, ctx.Err()
+	}
+}
+
+// Shutdown causes everything to clean up. We no longer need the scheduler.
+// TODO: should this be named Close() instead? Should it return an error?
+func (obj *Result) Shutdown() {
+	obj.closeFunc()
+	// XXX: should we have a waitgroup to wait for it all to close?
+}
+
+// Schedule returns a scheduler result which can be queried with it's available
+// methods. This automatically causes different etcd clients sharing the same
+// path to discover each other and be part of the scheduled set. On close the
+// keys expire and will get removed from the scheduled set. Different options
+// can be passed in to customize the behaviour. Hostname represents the unique
+// identifier for the caller. The behaviour is undefined if this is run more
+// than once with the same path and hostname simultaneously.
+func Schedule(client *etcd.Client, path string, hostname string, opts ...Option) (*Result, error) {
+	if strings.HasSuffix(path, "/") {
+		return nil, fmt.Errorf("scheduler: path must not end with the slash char")
+	}
+	if !strings.HasPrefix(path, "/") {
+		return nil, fmt.Errorf("scheduler: path must start with the slash char")
+	}
+	if hostname == "" {
+		return nil, fmt.Errorf("scheduler: hostname must not be empty")
+	}
+	if strings.Contains(hostname, hostnameJoinChar) {
+		return nil, fmt.Errorf("scheduler: hostname must not contain join char: %s", hostnameJoinChar)
+	}
+
+	// key structure is $path/election = ???
+	// key structure is $path/exchange/$hostname = ???
+	// key structure is $path/scheduled = ???
+
+	options := &schedulerOptions{ // default scheduler options
+		// If reuseLease is false, then on host disconnect, that hosts
+		// entry will immediately expire, and the scheduler will react
+		// instantly and remove that host entry from the list. If this
+		// is true, or if the host closes without a clean shutdown, it
+		// will take the TTL number of seconds to remove the key. This
+		// can be set using the concurrency.WithTTL option to Session.
+		reuseLease: false,
+		sessionTTL: DefaultSessionTTL,
+		maxCount:   DefaultMaxCount,
+	}
+	for _, optionFunc := range opts { // apply the scheduler options
+		optionFunc(options)
+	}
+
+	if options.strategy == nil {
+		return nil, fmt.Errorf("scheduler: strategy must be specified")
+	}
+
+	sessionOptions := []concurrency.SessionOption{}
+
+	// here we try to re-use lease between multiple runs of the code
+	// TODO: is it a good idea to try and re-use the lease b/w runs?
+	if options.reuseLease {
+		if leaseID, exists := schedulerLeases[path]; exists {
+			sessionOptions = append(sessionOptions, concurrency.WithLease(leaseID))
+		}
+	}
+	// ttl for key expiry on abrupt disconnection or if reuseLease is true!
+	if options.sessionTTL > 0 {
+		sessionOptions = append(sessionOptions, concurrency.WithTTL(options.sessionTTL))
+	}
+
+	//options.debug = true // use this for local debugging
+	session, err := concurrency.NewSession(client, sessionOptions...)
+	if err != nil {
+		return nil, errwrap.Wrapf(err, "scheduler: could not create session")
+	}
+	leaseID := session.Lease()
+	if options.reuseLease {
+		// save for next time, otherwise run session.Close() somewhere
+		schedulerLeases[path] = leaseID
+	}
+
+	ctx, cancel := context.WithCancel(context.Background()) // cancel below
+	//defer cancel() // do NOT do this, as it would cause an early cancel!
+
+	// stored scheduler results
+	scheduledPath := fmt.Sprintf("%s/scheduled", path)
+	scheduledChan := client.Watcher.Watch(ctx, scheduledPath)
+
+	// exchange hostname, and attach it to session (leaseID) so it expires
+	// (gets deleted) when we disconnect...
+	exchangePath := fmt.Sprintf("%s/exchange", path)
+	exchangePathHost := fmt.Sprintf("%s/%s", exchangePath, hostname)
+	exchangePathPrefix := fmt.Sprintf("%s/", exchangePath)
+
+	// open the watch *before* we set our key so that we can see the change!
+	watchChan := client.Watcher.Watch(ctx, exchangePathPrefix, etcd.WithPrefix())
+
+	data := "TODO" // XXX: no data to exchange alongside hostnames yet
+	ifops := []etcd.Cmp{
+		etcd.Compare(etcd.Value(exchangePathHost), "=", data),
+		etcd.Compare(etcd.LeaseValue(exchangePathHost), "=", leaseID),
+	}
+	elsop := etcd.OpPut(exchangePathHost, data, etcd.WithLease(leaseID))
+
+	// it's important to do this in one transaction, and atomically, because
+	// this way, we only generate one watch event, and only when it's needed
+	// updating leaseID, or key expiry (deletion) both generate watch events
+	// XXX: context!!!
+	if txn, err := client.KV.Txn(context.TODO()).If(ifops...).Then([]etcd.Op{}...).Else(elsop).Commit(); err != nil {
+		defer cancel() // cancel to avoid leaks if we exit early...
+		return nil, errwrap.Wrapf(err, "could not exchange in `%s`", path)
+	} else if txn.Succeeded {
+		options.logf("txn did nothing...") // then branch
+	} else {
+		options.logf("txn did an update...")
+	}
+
+	// create an election object
+	electionPath := fmt.Sprintf("%s/election", path)
+	election := concurrency.NewElection(session, electionPath)
+	electionChan := election.Observe(ctx)
+
+	elected := "" // who we "assume" is elected
+	wg := &sync.WaitGroup{}
+	ch := make(chan *schedulerResult)
+	closeChan := make(chan struct{})
+	send := func(hosts []string, err error) bool { // helper function for sending
+		select {
+		case ch <- &schedulerResult{ // send
+			hosts: hosts,
+			err:   err,
+		}:
+			return true
+		case <-closeChan: // unblock
+			return false // not sent
+		}
+	}
+
+	once := &sync.Once{}
+	onceBody := func() { // do not call directly, use closeFunc!
+		//cancel() // TODO: is this needed here?
+		// request a graceful shutdown, caller must call this to
+		// shutdown when they are finished with the scheduler...
+		// calling this will cause their hosts channels to close
+		close(closeChan) // send a close signal
+	}
+	closeFunc := func() {
+		once.Do(onceBody)
+	}
+	result := &Result{
+		results: ch,
+		// TODO: we could accept a context to watch for cancel instead?
+		closeFunc: closeFunc,
+	}
+
+	mutex := &sync.Mutex{}
+	var campaignClose chan struct{}
+	var campaignRunning bool
+	// goroutine to vote for someone as scheduler! each participant must be
+	// able to run this or nobody will be around to vote if others are down
+	campaignFunc := func() {
+		options.logf("starting campaign...")
+		// the mutex ensures we don't fly past the wg.Wait() if someone
+		// shuts down the scheduler right as we are about to start this
+		// campaigning loop up. we do not want to fail unnecessarily...
+		mutex.Lock()
+		wg.Add(1)
+		mutex.Unlock()
+		go func() {
+			defer wg.Done()
+
+			ctx, cancel := context.WithCancel(context.Background())
+			go func() {
+				defer cancel() // run cancel to stop campaigning...
+				select {
+				case <-campaignClose:
+					return
+				case <-closeChan:
+					return
+				}
+			}()
+			for {
+				// TODO: previously, this looped infinitely fast
+				// TODO: add some rate limiting here for initial
+				// campaigning which occasionally loops a lot...
+				if options.debug {
+					//fmt.Printf(".") // debug
+					options.logf("campaigning...")
+				}
+
+				// "Campaign puts a value as eligible for the election.
+				// It blocks until it is elected, an error occurs, or
+				// the context is cancelled."
+
+				// vote for ourselves, as it's the only host we can
+				// guarantee is alive, otherwise we wouldn't be voting!
+				// it would be more sensible to vote for the last valid
+				// hostname to keep things more stable, but if that
+				// information was stale, and that host wasn't alive,
+				// then this would defeat the point of picking them!
+				if err := election.Campaign(ctx, hostname); err != nil {
+					if err != context.Canceled {
+						send(nil, errwrap.Wrapf(err, "scheduler: error campaigning"))
+					}
+					return
+				}
+			}
+		}()
+	}
+
+	go func() {
+		defer close(ch)
+		if !options.reuseLease {
+			defer session.Close() // this revokes the lease...
+		}
+
+		defer func() {
+			// XXX: should we ever resign? why would this block and thus need a context?
+			if elected == hostname { // TODO: is it safe to just always do this?
+				if err := election.Resign(context.TODO()); err != nil { // XXX: add a timeout?
+				}
+			}
+			elected = "" // we don't care anymore!
+		}()
+
+		// this "last" defer (first to run) should block until the other
+		// goroutine has closed so we don't Close an in-use session, etc
+		defer wg.Wait()
+
+		go func() {
+			defer cancel() // run cancel to "free" Observe...
+
+			defer wg.Wait() // also wait here if parent exits first
+
+			select {
+			case <-closeChan:
+				// we want the above wg.Wait() to work if this
+				// close happens. lock with the campaign start
+				defer mutex.Unlock()
+				mutex.Lock()
+				return
+			}
+		}()
+		hostnames := make(map[string]string)
+		for {
+			select {
+			case val, ok := <-electionChan:
+				if options.debug {
+					options.logf("electionChan(%t): %+v", ok, val)
+				}
+				if !ok {
+					if options.debug {
+						options.logf("elections stream shutdown...")
+					}
+					electionChan = nil
+					// done
+					// TODO: do we need to send on error channel?
+					// XXX: maybe if context was not called to exit us?
+
+					// ensure everyone waiting on closeChan
+					// gets cleaned up so we free mem, etc!
+					if watchChan == nil && scheduledChan == nil { // all now closed
+						closeFunc()
+						return
+					}
+					continue
+
+				}
+
+				elected = string(val.Kvs[0].Value)
+				//if options.debug {
+				options.logf("elected: %s", elected)
+				//}
+				if elected != hostname { // not me!
+					// start up the campaign function
+					if !campaignRunning {
+						campaignClose = make(chan struct{})
+						campaignFunc() // run
+						campaignRunning = true
+					}
+					continue // someone else does the scheduling...
+				} else { // campaigning while i am it loops fast
+					// shutdown the campaign function
+					if campaignRunning {
+						close(campaignClose)
+						wg.Wait()
+						campaignRunning = false
+					}
+				}
+
+				// i was voted in to make the scheduling choice!
+
+			case watchResp, ok := <-watchChan:
+				if options.debug {
+					options.logf("watchChan(%t): %+v", ok, watchResp)
+				}
+				if !ok {
+					if options.debug {
+						options.logf("watch stream shutdown...")
+					}
+					watchChan = nil
+					// done
+					// TODO: do we need to send on error channel?
+					// XXX: maybe if context was not called to exit us?
+
+					// ensure everyone waiting on closeChan
+					// gets cleaned up so we free mem, etc!
+					if electionChan == nil && scheduledChan == nil { // all now closed
+						closeFunc()
+						return
+					}
+					continue
+				}
+
+				err := watchResp.Err()
+				if watchResp.Canceled || err == context.Canceled {
+					// channel get closed shortly...
+					continue
+				}
+				if watchResp.Header.Revision == 0 { // by inspection
+					// received empty message ?
+					// switched client connection ?
+					// FIXME: what should we do here ?
+					continue
+				}
+				if err != nil {
+					send(nil, errwrap.Wrapf(err, "scheduler: exchange watcher failed"))
+					continue
+				}
+				if len(watchResp.Events) == 0 { // nothing interesting
+					continue
+				}
+
+				options.logf("running exchange values get...")
+				resp, err := client.Get(ctx, exchangePathPrefix, etcd.WithPrefix(), etcd.WithSort(etcd.SortByKey, etcd.SortAscend))
+				if err != nil || resp == nil {
+					if err != nil {
+						send(nil, errwrap.Wrapf(err, "scheduler: could not get exchange values in `%s`", path))
+					} else { // if resp == nil
+						send(nil, fmt.Errorf("scheduler: could not get exchange values in `%s`, resp is nil", path))
+					}
+					continue
+				}
+
+				// FIXME: the value key could instead be host
+				// specific information which is used for some
+				// purpose, eg: seconds active, and other data?
+				hostnames = make(map[string]string) // reset
+				for _, x := range resp.Kvs {
+					k := string(x.Key)
+					if !strings.HasPrefix(k, exchangePathPrefix) {
+						continue
+					}
+					k = k[len(exchangePathPrefix):] // strip
+					hostnames[k] = string(x.Value)
+				}
+				if options.debug {
+					options.logf("available hostnames: %+v", hostnames)
+				}
+
+			case scheduledResp, ok := <-scheduledChan:
+				if options.debug {
+					options.logf("scheduledChan(%t): %+v", ok, scheduledResp)
+				}
+				if !ok {
+					if options.debug {
+						options.logf("scheduled stream shutdown...")
+					}
+					scheduledChan = nil
+					// done
+					// TODO: do we need to send on error channel?
+					// XXX: maybe if context was not called to exit us?
+
+					// ensure everyone waiting on closeChan
+					// gets cleaned up so we free mem, etc!
+					if electionChan == nil && watchChan == nil { // all now closed
+						closeFunc()
+						return
+					}
+					continue
+				}
+				// event! continue below and get new result...
+
+				// NOTE: not needed, exit this via Observe ctx cancel,
+				// which will ultimately cause the chan to shutdown...
+				//case <-closeChan:
+				//	return
+			} // end select
+
+			if len(hostnames) == 0 {
+				if options.debug {
+					options.logf("zero hosts available")
+				}
+				continue // not enough hosts available
+			}
+
+			// if we're currently elected, make a scheduling decision
+			// if not, lookup the existing leader scheduling decision
+			if elected != hostname {
+				options.logf("i am not the leader, running scheduling result get...")
+				resp, err := client.Get(ctx, scheduledPath)
+				if err != nil || resp == nil || len(resp.Kvs) != 1 {
+					if err != nil {
+						send(nil, errwrap.Wrapf(err, "scheduler: could not get scheduling result in `%s`", path))
+					} else if resp == nil {
+						send(nil, fmt.Errorf("scheduler: could not get scheduling result in `%s`, resp is nil", path))
+					} else if len(resp.Kvs) > 1 {
+						send(nil, fmt.Errorf("scheduler: could not get scheduling result in `%s`, resp kvs: %+v", path, resp.Kvs))
+					}
+					// if len(resp.Kvs) == 0, we shouldn't error
+					// in that situation it's just too early...
+					continue
+				}
+
+				result := string(resp.Kvs[0].Value)
+				hosts := strings.Split(result, hostnameJoinChar)
+
+				if options.debug {
+					options.logf("sending hosts: %+v", hosts)
+				}
+				// send that on channel!
+				if !send(hosts, nil) {
+					//return // pass instead, let channels clean up
+				}
+				continue
+			}
+
+			// i am the leader, run scheduler and store result
+			options.logf("i am elected, running scheduler...")
+
+			// run actual scheduler and decide who should be chosen
+			// TODO: is there any additional data that we can pass
+			// to the scheduler so it can make a better decision ?
+			hosts, err := options.strategy.Schedule(hostnames, options)
+			if err != nil {
+				send(nil, errwrap.Wrapf(err, "scheduler: strategy failed"))
+				continue
+			}
+			sort.Strings(hosts) // for consistency
+
+			options.logf("storing scheduling result...")
+			data := strings.Join(hosts, hostnameJoinChar)
+			ifops := []etcd.Cmp{
+				etcd.Compare(etcd.Value(scheduledPath), "=", data),
+			}
+			elsop := etcd.OpPut(scheduledPath, data)
+
+			// it's important to do this in one transaction, and atomically, because
+			// this way, we only generate one watch event, and only when it's needed
+			// updating leaseID, or key expiry (deletion) both generate watch events
+			// XXX: context!!!
+			if _, err := client.KV.Txn(context.TODO()).If(ifops...).Then([]etcd.Op{}...).Else(elsop).Commit(); err != nil {
+				send(nil, errwrap.Wrapf(err, "scheduler: could not set scheduling result in `%s`", path))
+				continue
+			}
+
+			if options.debug {
+				options.logf("sending hosts: %+v", hosts)
+			}
+			// send that on channel!
+			if !send(hosts, nil) {
+				//return // pass instead, let channels clean up
+			}
+		}
+	}()
+
+	// kick off an initial campaign if none exist already...
+	options.logf("checking for existing leader...")
+	leaderResult, err := election.Leader(ctx)
+	if err == concurrency.ErrElectionNoLeader {
+		// start up the campaign function
+		if !campaignRunning {
+			campaignClose = make(chan struct{})
+			campaignFunc() // run
+			campaignRunning = true
+		}
+	}
+	if options.debug {
+		if err != nil {
+			options.logf("leader information error: %+v", err)
+		} else {
+			options.logf("leader information: %+v", leaderResult)
+		}
+	}
+
+	return result, nil
+}

etcd/scheduler/strategy.go

@@ -0,0 +1,51 @@
+// Mgmt
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
+// Written by James Shubin <james@shubin.ca> and the project contributors
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+package scheduler
+
+import (
+	"fmt"
+)
+
+// registeredStrategies is a global map of all possible strategy implementations
+// which can be used. You should never touch this map directly. Use methods like
+// Register instead.
+var registeredStrategies = make(map[string]func() Strategy) // must initialize
+
+// Strategy represents the methods a scheduler strategy must implement.
+type Strategy interface {
+	Schedule(hostnames map[string]string, opts *schedulerOptions) ([]string, error)
+}
+
+// Register takes a func and its name and makes it available for use. It is
+// commonly called in the init() method of the func at program startup. There is
+// no matching Unregister function.
+func Register(name string, fn func() Strategy) {
+	if _, ok := registeredStrategies[name]; ok {
+		panic(fmt.Sprintf("a strategy named %s is already registered", name))
+	}
+	//gob.Register(fn())
+	registeredStrategies[name] = fn
+}
+
+type nilStrategy struct {
+}
+
+// Schedule returns an error for any scheduling request for this nil strategy.
+func (obj *nilStrategy) Schedule(hostnames map[string]string, opts *schedulerOptions) ([]string, error) {
+	return nil, fmt.Errorf("scheduler: cannot schedule with nil scheduler")
+}

etcd/str.go

@@ -1,5 +1,5 @@
 // Mgmt
-// Copyright (C) 2013-2017+ James Shubin and the project contributors
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
 // Written by James Shubin <james@shubin.ca> and the project contributors
 //
 // This program is free software: you can redistribute it and/or modify
@@ -32,9 +32,13 @@ var ErrNotExist = errors.New("errNotExist")
 // WatchStr returns a channel which spits out events on key activity.
 // FIXME: It should close the channel when it's done, and spit out errors when
 // something goes wrong.
+// XXX: since the caller of this (via the World API) has no way to tell it it's
+// done, does that mean we leak go-routines since it might still be running, but
+// perhaps even blocked??? Could this cause a dead-lock? Should we instead return
+// some sort of struct which has a close method with it to ask for a shutdown?
 func WatchStr(obj *EmbdEtcd, key string) chan error {
-	// new key structure is /$NS/strings/$key = $data
-	path := fmt.Sprintf("/%s/strings/%s", NS, key)
+	// new key structure is $NS/strings/$key = $data
+	path := fmt.Sprintf("%s/strings/%s", NS, key)
 	ch := make(chan error, 1)
 	// FIXME: fix our API so that we get a close event on shutdown.
 	callback := func(re *RE) error {
@@ -54,8 +58,8 @@ func WatchStr(obj *EmbdEtcd, key string) chan error {
 
 // GetStr collects the string which matches a global namespace in etcd.
 func GetStr(obj *EmbdEtcd, key string) (string, error) {
-	// new key structure is /$NS/strings/$key = $data
-	path := fmt.Sprintf("/%s/strings/%s", NS, key)
+	// new key structure is $NS/strings/$key = $data
+	path := fmt.Sprintf("%s/strings/%s", NS, key)
 	keyMap, err := obj.Get(path, etcd.WithPrefix())
 	if err != nil {
 		return "", errwrap.Wrapf(err, "could not get strings in: %s", key)
@@ -82,8 +86,8 @@ func GetStr(obj *EmbdEtcd, key string) (string, error) {
 // nil, then it deletes the key. Otherwise the value should point to a string.
 // TODO: TTL or delete disconnect?
 func SetStr(obj *EmbdEtcd, key string, data *string) error {
-	// key structure is /$NS/strings/$key = $data
-	path := fmt.Sprintf("/%s/strings/%s", NS, key)
+	// key structure is $NS/strings/$key = $data
+	path := fmt.Sprintf("%s/strings/%s", NS, key)
 	ifs := []etcd.Cmp{} // list matching the desired state
 	ops := []etcd.Op{}  // list of ops in this transaction (then)
 	els := []etcd.Op{}  // list of ops in this transaction (else)

etcd/strmap.go

@@ -1,5 +1,5 @@
 // Mgmt
-// Copyright (C) 2013-2017+ James Shubin and the project contributors
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
 // Written by James Shubin <james@shubin.ca> and the project contributors
 //
 // This program is free software: you can redistribute it and/or modify
@@ -31,8 +31,8 @@ import (
 // FIXME: It should close the channel when it's done, and spit out errors when
 // something goes wrong.
 func WatchStrMap(obj *EmbdEtcd, key string) chan error {
-	// new key structure is /$NS/strings/$key/$hostname = $data
-	path := fmt.Sprintf("/%s/strings/%s", NS, key)
+	// new key structure is $NS/strings/$key/$hostname = $data
+	path := fmt.Sprintf("%s/strings/%s", NS, key)
 	ch := make(chan error, 1)
 	// FIXME: fix our API so that we get a close event on shutdown.
 	callback := func(re *RE) error {
@@ -52,12 +52,12 @@ func WatchStrMap(obj *EmbdEtcd, key string) chan error {
 
 // GetStrMap collects all of the strings which match a namespace in etcd.
 func GetStrMap(obj *EmbdEtcd, hostnameFilter []string, key string) (map[string]string, error) {
-	// old key structure is /$NS/strings/$hostname/$key = $data
-	// new key structure is /$NS/strings/$key/$hostname = $data
+	// old key structure is $NS/strings/$hostname/$key = $data
+	// new key structure is $NS/strings/$key/$hostname = $data
 	// FIXME: if we have the $key as the last token (old key structure), we
 	// can allow the key to contain the slash char, otherwise we need to
 	// verify that one isn't present in the input string.
-	path := fmt.Sprintf("/%s/strings/%s", NS, key)
+	path := fmt.Sprintf("%s/strings/%s", NS, key)
 	keyMap, err := obj.Get(path, etcd.WithPrefix(), etcd.WithSort(etcd.SortByKey, etcd.SortAscend))
 	if err != nil {
 		return nil, errwrap.Wrapf(err, "could not get strings in: %s", key)
@@ -92,8 +92,8 @@ func GetStrMap(obj *EmbdEtcd, hostnameFilter []string, key string) (map[string]s
 // nil, then it deletes the key. Otherwise the value should point to a string.
 // TODO: TTL or delete disconnect?
 func SetStrMap(obj *EmbdEtcd, hostname, key string, data *string) error {
-	// key structure is /$NS/strings/$key/$hostname = $data
-	path := fmt.Sprintf("/%s/strings/%s/%s", NS, key, hostname)
+	// key structure is $NS/strings/$key/$hostname = $data
+	path := fmt.Sprintf("%s/strings/%s/%s", NS, key, hostname)
 	ifs := []etcd.Cmp{} // list matching the desired state
 	ops := []etcd.Op{}  // list of ops in this transaction (then)
 	els := []etcd.Op{}  // list of ops in this transaction (else)

etcd/world.go

@@ -1,5 +1,5 @@
 // Mgmt
-// Copyright (C) 2013-2017+ James Shubin and the project contributors
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
 // Written by James Shubin <james@shubin.ca> and the project contributors
 //
 // This program is free software: you can redistribute it and/or modify
@@ -18,13 +18,24 @@
 package etcd
 
 import (
+	"fmt"
+	"net/url"
+	"strings"
+
+	etcdfs "github.com/purpleidea/mgmt/etcd/fs"
+	"github.com/purpleidea/mgmt/etcd/scheduler"
 	"github.com/purpleidea/mgmt/resources"
 )
 
 // World is an etcd backed implementation of the World interface.
 type World struct {
-	Hostname string // uuid for the consumer of these
-	EmbdEtcd *EmbdEtcd
+	Hostname       string // uuid for the consumer of these
+	EmbdEtcd       *EmbdEtcd
+	MetadataPrefix string       // expected metadata prefix
+	StoragePrefix  string       // storage prefix for etcdfs storage
+	StandaloneFs   resources.Fs // store an fs here for local usage
+	Debug          bool
+	Logf           func(format string, v ...interface{})
 }
 
 // ResWatch returns a channel which spits out events on possible exported
@@ -93,3 +104,49 @@ func (obj *World) StrMapSet(namespace, value string) error {
 func (obj *World) StrMapDel(namespace string) error {
 	return SetStrMap(obj.EmbdEtcd, obj.Hostname, namespace, nil)
 }
+
+// Scheduler returns a scheduling result of hosts in a particular namespace.
+func (obj *World) Scheduler(namespace string, opts ...scheduler.Option) (*scheduler.Result, error) {
+	modifiedOpts := []scheduler.Option{}
+	for _, o := range opts {
+		modifiedOpts = append(modifiedOpts, o) // copy in
+	}
+
+	modifiedOpts = append(modifiedOpts, scheduler.Debug(obj.Debug))
+	modifiedOpts = append(modifiedOpts, scheduler.Logf(obj.Logf))
+
+	return scheduler.Schedule(obj.EmbdEtcd.GetClient(), fmt.Sprintf("%s/scheduler/%s", NS, namespace), obj.Hostname, modifiedOpts...)
+}
+
+// Fs returns a distributed file system from a unique URI. For single host
+// execution that doesn't span more than a single host, this file system might
+// actually be a local or memory backed file system, so actually only
+// distributed within the boredom that is a single host cluster.
+func (obj *World) Fs(uri string) (resources.Fs, error) {
+	u, err := url.Parse(uri)
+	if err != nil {
+		return nil, err
+	}
+
+	// we're in standalone mode
+	if u.Scheme == "memmapfs" && u.Path == "/" {
+		return obj.StandaloneFs, nil
+	}
+
+	if u.Scheme != "etcdfs" {
+		return nil, fmt.Errorf("unknown scheme: `%s`", u.Scheme)
+	}
+	if u.Path == "" {
+		return nil, fmt.Errorf("empty path: %s", u.Path)
+	}
+	if !strings.HasPrefix(u.Path, obj.MetadataPrefix) {
+		return nil, fmt.Errorf("wrong path prefix: %s", u.Path)
+	}
+
+	etcdFs := &etcdfs.Fs{
+		Client:     obj.EmbdEtcd.GetClient(),
+		Metadata:   u.Path,
+		DataPrefix: obj.StoragePrefix,
+	}
+	return etcdFs, nil
+}

event/event.go

@@ -1,5 +1,5 @@
 // Mgmt
-// Copyright (C) 2013-2017+ James Shubin and the project contributors
+// Copyright (C) 2013-2018+ James Shubin and the project contributors
 // Written by James Shubin <james@shubin.ca> and the project contributors
 //
 // This program is free software: you can redistribute it and/or modify

examples/lang/answer.mcl

@@ -0,0 +1,4 @@
+# it was a lovely surprise to me, when i realized that mgmt had the answer!
+print "answer" {
+	msg => printf("the answer to life, the universe, and everything is: %d", answer()),
+}

examples/lang/contains0.mcl

@@ -0,0 +1,22 @@
+$set = ["a", "b", "c", "d",]
+
+$c1 = "x1" in ["x1", "x2", "x3",]
+$c2 = 42 in [4, 13, 42,]
+$c3 = "x" in $set
+$c4 = "b" in $set
+
+$s = printf("1: %t, 2: %t, 3: %t, 4: %t\n", $c1, $c2, $c3, $c4)
+
+file "/tmp/mgmt/contains" {
+	content => $s,
+}
+
+$x = if hostname() in ["h1", "h3",] {
+	printf("i (%s) am one of the chosen few!\n", hostname())
+} else {
+	printf("i (%s) was not chosen :(\n", hostname())
+}
+
+file "/tmp/mgmt/hello-${hostname()}" {
+	content => $x,
+}

examples/lang/datetime1.mcl

@@ -0,0 +1,4 @@
+$d = datetime()
+file "/tmp/mgmt/datetime" {
+	content => template("Hello! It is now: {{ datetime_print . }}\n", $d),
+}

examples/lang/datetime2.mcl

@@ -0,0 +1,14 @@
+$secplusone = datetime() + $ayear
+
+# note the order of the assignment (year can come later in the code)
+$ayear = 60 * 60 * 24 * 365	# is a year in seconds (31536000)
+
+$tmplvalues = struct{year => $secplusone, load => $theload,}
+
+$theload = structlookup(load(), "x1")
+
+if 5 > 3 {
+	file "/tmp/mgmt/datetime" {
+		content => template("Now + 1 year is: {{ .year }} seconds, aka: {{ datetime_print .year }}\n\nload average: {{ .load }}\n", $tmplvalues),
+	}
+}

examples/lang/datetime3.mcl

@@ -0,0 +1,14 @@
+$secplusone = datetime() + $ayear
+
+# note the order of the assignment (year can come later in the code)
+$ayear = 60 * 60 * 24 * 365	# is a year in seconds (31536000)
+
+$tmplvalues = struct{year => $secplusone, load => $theload, vumeter => $vumeter,}
+
+$theload = structlookup(load(), "x1")
+
+$vumeter = vumeter("====", 10, 0.9)
+
+file "/tmp/mgmt/datetime" {
+	content => template("Now + 1 year is: {{ .year }} seconds, aka: {{ datetime_print .year }}\n\nload average: {{ .load }}\n\nvu: {{ .vumeter }}\n", $tmplvalues),
+}

examples/lang/edges0.mcl

@@ -0,0 +1,14 @@
+exec "exec0" {
+	cmd => "sleep 10s",
+	shell => "/bin/bash",
+}
+exec "exec1" {
+	cmd => "sleep 10s",
+	shell => "/bin/bash",
+}
+exec "exec2" {
+	cmd => "sleep 10s",
+	shell => "/bin/bash",
+}
+
+Exec["exec0"] -> Exec["exec1"] -> Exec["exec2"]

examples/lang/edges1.mcl

@@ -0,0 +1,18 @@
+$b = true
+if $b {
+	exec "exec0" {
+		cmd => "sleep 10s",
+		shell => "/bin/bash",
+	}
+}
+exec "exec1" {
+	cmd => "sleep 10s",
+	shell => "/bin/bash",
+
+	Depend => $b ?: Exec["exec0"],
+	Before => Exec["exec2"],
+}
+exec "exec2" {
+	cmd => "sleep 10s",
+	shell => "/bin/bash",
+}

examples/lang/elvis0.mcl

@@ -0,0 +1,5 @@
+$b = true # change me to false and then try editing the file manually
+file "/tmp/mgmt-elvis" {
+	content => $b ?: "hello world\n",
+	state => "exists",
+}

examples/lang/env0.mcl

@@ -0,0 +1,20 @@
+# read and print environment variable
+# env TEST=123 EMPTY= ./mgmt run --tmp-prefix --lang=examples/lang/env0.mcl --converged-timeout=5
+
+$x = getenv("TEST", "321")
+
+print "print1" {
+	msg => printf("the value of the environment variable TEST is: %s", $x),
+}
+
+$y = getenv("DOESNOTEXIT", "321")
+
+print "print2" {
+	msg => printf("environment variable DOESNOTEXIT does not exist, defaulting to: %s", $y),
+}
+
+$z = getenv("EMPTY", "456")
+
+print "print3" {
+	msg => printf("same goes for epmty variables like EMPTY: %s", $z),
+}

examples/lang/env1.mcl

@@ -0,0 +1,10 @@
+$env = env()
+$m = maplookup($env, "GOPATH", "")
+
+print "print0" {
+	msg => if hasenv("GOPATH") {
+		printf("GOPATH is: %s", $m)
+	} else {
+		"GOPATH is missing!"
+	},
+}

examples/lang/exchange0.mcl

@@ -0,0 +1,13 @@
+# run this example with these commands
+# watch -n 0.1 'tail *'	# run this in /tmp/mgmt/
+# time ./mgmt run --lang examples/lang/exchange0.mcl --hostname h1 --ideal-cluster-size 1 --tmp-prefix --no-pgp
+# time ./mgmt run --lang examples/lang/exchange0.mcl --hostname h2 --seeds http://127.0.0.1:2379 --client-urls http://127.0.0.1:2381 --server-urls http://127.0.0.1:2382 --tmp-prefix --no-pgp
+# time ./mgmt run --lang examples/lang/exchange0.mcl --hostname h3 --seeds http://127.0.0.1:2379 --client-urls http://127.0.0.1:2383 --server-urls http://127.0.0.1:2384 --tmp-prefix --no-pgp
+# time ./mgmt run --lang examples/lang/exchange0.mcl --hostname h4 --seeds http://127.0.0.1:2379 --client-urls http://127.0.0.1:2385 --server-urls http://127.0.0.1:2386 --tmp-prefix --no-pgp
+
+$rand = random1(8)
+$exchanged = exchange("keyns", $rand)
+
+file "/tmp/mgmt/exchange-${hostname()}" {
+	content => template("Found: {{ . }}\n", $exchanged),
+}

examples/lang/hello0.mcl

@@ -0,0 +1,4 @@
+file "/tmp/mgmt-hello-world" {
+	content => "hello world from @purpleidea\n",
+	state => "exists",
+}

examples/lang/history1.mcl

@@ -0,0 +1,7 @@
+$dt = datetime()
+
+$hystvalues = {"ix0" => $dt, "ix1" => $dt{1}, "ix2" => $dt{2}, "ix3" => $dt{3},}
+
+file "/tmp/mgmt/history" {
+	content => template("Index(0) {{.ix0}}: {{ datetime_print .ix0 }}\nIndex(1) {{.ix1}}: {{ datetime_print .ix1 }}\nIndex(2) {{.ix2}}: {{ datetime_print .ix2 }}\nIndex(3) {{.ix3}}: {{ datetime_print .ix3 }}\n", $hystvalues),
+}

examples/lang/hostname0.mcl

@@ -0,0 +1,4 @@
+file "/tmp/mgmt/${hostname()}" {
+	content => "hello from ${hostname()}!\n",
+	state => "exists",
+}

examples/lang/hysteresis1.mcl

@@ -0,0 +1,31 @@
+file "/tmp/mgmt/systemload" {
+	content => template("load average: {{ .load }} threshold: {{ .threshold }}\n", $tmplvalues),
+}
+
+$tmplvalues = struct{load => $theload, threshold => $threshold,}
+
+$theload = structlookup(load(), "x1")
+$threshold = 1.5	# change me if you like
+
+# simple hysteresis implementation
+$h1 = $theload > $threshold
+$h2 = $theload{1} > $threshold
+$h3 = $theload{2} > $threshold
+$unload = $h1 || $h2 || $h3
+
+virt "mgmt1" {
+	uri => "qemu:///session",
+	cpus => 1,
+	memory => 524288,
+	state => "running",
+	transient => true,
+}
+
+# this vm shuts down under load...
+virt "mgmt2" {
+	uri => "qemu:///session",
+	cpus => 1,
+	memory => 524288,
+	state => if $unload { "shutoff" } else { "running" },
+	transient => true,
+}

examples/lang/interpolate1.mcl

@@ -0,0 +1,5 @@
+$audience = "WORLD!"
+file "/tmp/mgmt/hello" {
+	content => "hello ${audience}!\n",
+	state => "exists",
+}

examples/lang/len0.mcl

@@ -0,0 +1,9 @@
+$x1 = ["a", "b", "c", "d",]
+print "print4" {
+	msg => printf("length is: %d", len($x1)),
+}
+
+$x2 = {"a" => 1, "b" => 2, "c" => 3,}
+print "print3" {
+	msg => printf("length is: %d", len($x2)),
+}

examples/lang/load0.mcl

@@ -0,0 +1,9 @@
+$theload = load()
+
+$x1 = structlookup($theload, "x1")
+$x5 = structlookup($theload, "x5")
+$x15 = structlookup($theload, "x15")
+
+print "print1" {
+	msg => printf("load average: %f, %f, %f", $x1, $x5, $x15),
+}

examples/lang/maplookup1.mcl

@@ -0,0 +1,13 @@
+$m = {"k1" => 42, "k2" => 13,}
+
+$found = maplookup($m, "k1", 99)
+
+print "print1" {
+	msg => printf("found value of: %d", $found),
+}
+
+$notfound = maplookup($m, "k3", 99)
+
+print "print2" {
+	msg => printf("notfound value of: %d", $notfound),
+}

examples/lang/math1.mcl

@@ -0,0 +1,4 @@
+test "t1" {
+	int64 => (4 + 32) * 15 - 8,
+	anotherstr => printf("the answer is: %d", 42),
+}

examples/lang/math2.mcl

@@ -0,0 +1,3 @@
+print "print0" {
+	msg => printf("13.0 ^ 4.2 is: %f", pow(13.0, 4.2)),
+}

examples/lang/password0.mcl

@@ -0,0 +1,8 @@
+password "pass0" {
+	length => 8,
+}
+
+file "/tmp/mgmt/password" {
+}
+
+Password["pass0"].password -> File["/tmp/mgmt/password"].content

examples/lang/pkg1.mcl

@@ -0,0 +1,3 @@
+pkg "cowsay" {
+	state => "installed",
+}

examples/lang/printf1.mcl

@@ -0,0 +1,8 @@
+test "printf-a" {
+	anotherstr => printf("the %s is: %d", "answer", 42),
+}
+
+$format = "a %s is: %f"
+test "printf-b" {
+	anotherstr => printf($format, "cool number", 3.14159),
+}

examples/lang/schedule0.mcl

@@ -0,0 +1,18 @@
+# here are all the possible options:
+#$opts = struct{strategy => "rr", max => 3, reuse => false, ttl => 10,}
+
+# although an empty struct is valid too:
+#$opts = struct{}
+
+# we'll just use a smaller subset today:
+$opts = struct{strategy => "rr", max => 2, ttl => 10,}
+
+# schedule in a particular namespace with options:
+$set = schedule("xsched", $opts)
+
+# and if you want, you can omit the options entirely:
+#$set = schedule("xsched")
+
+file "/tmp/mgmt/scheduled-${hostname()}" {
+	content => template("set: {{ . }}\n", $set),
+}

examples/lang/sendrecv0.mcl

@@ -0,0 +1,9 @@
+exec "exec0" {
+	cmd => "echo hello world && echo goodbye world 1>&2", # to stdout && stderr
+	shell => "/bin/bash",
+}
+
+print "print0" {
+}
+
+Exec["exec0"].output -> Print["print0"].msg

examples/lang/sendrecv1.mcl

@@ -0,0 +1,14 @@
+file "/tmp/mgmt/foo" {
+	content => "hello from foo\n",
+}
+
+print "print0" {
+}
+
+File["/tmp/mgmt/foo"].content -> Print["print0"].msg
+
+print "print1" {
+	msg => "hello",
+}
+
+Print["print0"] -> Print["print1"]

examples/lang/sendrecv2.mcl

@@ -0,0 +1,8 @@
+file "/tmp/mgmt/foo" {
+	content => "hello from foo\n",
+}
+
+file "/tmp/mgmt/bar" {
+}
+
+File["/tmp/mgmt/foo"].content -> File["/tmp/mgmt/bar"].content

examples/lang/sendrecv3.mcl

@@ -0,0 +1,21 @@
+$ns = "estate"
+$exchanged = kvlookup($ns)
+$state = maplookup($exchanged, $hostname, "default")
+
+exec "exec0" {
+	cmd => "echo hello world && echo goodbye world 1>&2", # to stdout && stderr
+	shell => "/bin/bash",
+}
+
+kv "kv0" {
+	key => $ns,
+	#value => "two",
+}
+
+Exec["exec0"].output -> Kv["kv0"].value
+
+if $state != "default" {
+	file "/tmp/mgmt/state" {
+		content => printf("state: %s\n", $state),
+	}
+}

examples/lang/shadowing1.mcl

@@ -0,0 +1,10 @@
+$x = "hello"
+if true {
+	$x = "i am shadowed"	# this is allowed, but not a good practice to intentionally shadow
+	print "inner-scope" {
+		msg => $x,	# contents are: i am shadowed
+	}
+}
+print "top-scope" {
+	msg => $x,	# contents are: hello
+}

examples/lang/states0.mcl

@@ -0,0 +1,50 @@
+$ns = "estate"
+$exchanged = kvlookup($ns)
+
+$state = maplookup($exchanged, $hostname, "default")
+
+if $state == "one" || $state == "default" {
+
+	file "/tmp/mgmt/state" {
+		content => "state: one\n",
+	}
+
+	exec "timer" {
+		cmd => "/usr/bin/sleep 1s",
+	}
+	kv "${ns}" {
+		key => $ns,
+		value => "two",
+	}
+	Exec["timer"] -> Kv["${ns}"]
+}
+if $state == "two" {
+
+	file "/tmp/mgmt/state" {
+		content => "state: two\n",
+	}
+
+	exec "timer" {
+		cmd => "/usr/bin/sleep 1s",
+	}
+	kv "${ns}" {
+		key => $ns,
+		value => "three",
+	}
+	Exec["timer"] -> Kv["${ns}"]
+}
+if $state == "three" {
+
+	file "/tmp/mgmt/state" {
+		content => "state: three\n",
+	}
+
+	exec "timer" {
+		cmd => "/usr/bin/sleep 1s",
+	}
+	kv "${ns}" {
+		key => $ns,
+		value => "one",
+	}
+	Exec["timer"] -> Kv["${ns}"]
+}

examples/lang/structlookup1.mcl

@@ -0,0 +1,13 @@
+$st = struct{f1 => 42, f2 => true, f3 => 3.14,}
+
+$f1 = structlookup($st, "f1")
+
+print "print1" {
+	msg => printf("f1 field is: %d", $f1),
+}
+
+$f2 = structlookup($st, "f2")
+
+print "print2" {
+	msg => printf("f2 field is: %t", $f2),
+}

examples/lang/template0.mcl

@@ -0,0 +1,10 @@
+$answer = 42
+$s = int2str($answer)
+
+print "print1" {
+	msg => printf("an str is: %s", $s),
+}
+
+print "print2" {
+	msg => template("an str is: {{ int2str . }}", $answer),
+}

examples/lang/virt1.mcl

@@ -0,0 +1,7 @@
+virt "mgmt3" {
+	uri => "qemu:///session",
+	cpus => 1,
+	memory => 524288,
+	state => "running",
+	transient => true,
+}

examples/lib/exec-send-recv.go

@@ -14,6 +14,15 @@ import (
 	mgmt "github.com/purpleidea/mgmt/lib"
 	"github.com/purpleidea/mgmt/pgraph"
 	"github.com/purpleidea/mgmt/resources"
+
+	"github.com/urfave/cli"
+)
+
+// XXX: this has not been updated to latest GAPI/Deploy API. Patches welcome!
+
+const (
+	// Name is the name of this frontend.
+	Name = "libmgmt"
 )
 
 // MyGAPI implements the main GAPI interface.
@@ -36,6 +45,39 @@ func NewMyGAPI(data gapi.Data, name string, interval uint) (*MyGAPI, error) {
 	return obj, obj.Init(data)
 }
 
+// Cli takes a cli.Context, and returns our GAPI if activated. All arguments
+// should take the prefix of the registered name. On activation, if there are
+// any validation problems, you should return an error. If this was not
+// activated, then you should return a nil GAPI and a nil error.
+func (obj *MyGAPI) Cli(c *cli.Context, fs resources.Fs) (*gapi.Deploy, error) {
+	if s := c.String(obj.Name); c.IsSet(obj.Name) {
+		if s != "" {
+			return nil, fmt.Errorf("input is not empty")
+		}
+
+		return &gapi.Deploy{
+			Name: obj.Name,
+			Noop: c.GlobalBool("noop"),
+			Sema: c.GlobalInt("sema"),
+			GAPI: &MyGAPI{
+			// TODO: add properties here...
+			},
+		}, nil
+	}
+	return nil, nil // we weren't activated!
+}
+
+// CliFlags returns a list of flags used by this deploy subcommand.
+func (obj *MyGAPI) CliFlags() []cli.Flag {
+	return []cli.Flag{
+		cli.StringFlag{
+			Name:  obj.Name,
+			Value: "",
+			Usage: "run",
+		},
+	}
+}
+
 // Init initializes the MyGAPI struct.
 func (obj *MyGAPI) Init(data gapi.Data) error {
 	if obj.initialized {
@@ -53,7 +95,7 @@ func (obj *MyGAPI) Init(data gapi.Data) error {
 // Graph returns a current Graph.
 func (obj *MyGAPI) Graph() (*pgraph.Graph, error) {
 	if !obj.initialized {
-		return nil, fmt.Errorf("libmgmt: MyGAPI is not initialized")
+		return nil, fmt.Errorf("%s: MyGAPI is not initialized", Name)
 	}
 
 	g, err := pgraph.NewGraph(obj.Name)
@@ -135,7 +177,7 @@ func (obj *MyGAPI) Next() chan gapi.Next {
 		defer close(ch) // this will run before the obj.wg.Done()
 		if !obj.initialized {
 			next := gapi.Next{
-				Err:  fmt.Errorf("libmgmt: MyGAPI is not initialized"),
+				Err:  fmt.Errorf("%s: MyGAPI is not initialized", Name),
 				Exit: true, // exit, b/c programming error?
 			}
 			ch <- next
@@ -164,7 +206,7 @@ func (obj *MyGAPI) Next() chan gapi.Next {
 				return
 			}
 
-			log.Printf("libmgmt: Generating new graph...")
+			log.Printf("%s: Generating new graph...", Name)
 			select {
 			case ch <- gapi.Next{}: // trigger a run
 			case <-obj.closeChan:
@@ -178,7 +220,7 @@ func (obj *MyGAPI) Next() chan gapi.Next {
 // Close shuts down the MyGAPI.
 func (obj *MyGAPI) Close() error {
 	if !obj.initialized {
-		return fmt.Errorf("libmgmt: MyGAPI is not initialized")
+		return fmt.Errorf("%s: MyGAPI is not initialized", Name)
 	}
 	close(obj.closeChan)
 	obj.wg.Wait()
@@ -199,10 +241,10 @@ func Run() error {
 	obj.ConvergedTimeout = -1
 	obj.Noop = false // FIXME: careful!
 
-	obj.GAPI = &MyGAPI{ // graph API
-		Name:     "libmgmt", // TODO: set on compilation
-		Interval: 60 * 10,   // arbitrarily change graph every 15 seconds
-	}
+	//obj.GAPI = &MyGAPI{ // graph API
+	//	Name:     "libmgmt", // TODO: set on compilation
+	//	Interval: 60 * 10,   // arbitrarily change graph every 15 seconds
+	//}
 
 	if err := obj.Init(); err != nil {
 		return err

examples/lib/libmgmt-subgraph0.go

@@ -16,8 +16,20 @@ import (
 	"github.com/purpleidea/mgmt/resources"
 
 	errwrap "github.com/pkg/errors"
+	"github.com/urfave/cli"
 )
 
+// XXX: this has not been updated to latest GAPI/Deploy API. Patches welcome!
+
+const (
+	// Name is the name of this frontend.
+	Name = "libmgmt"
+)
+
+func init() {
+	gapi.Register(Name, func() gapi.GAPI { return &MyGAPI{} }) // register
+}
+
 // MyGAPI implements the main GAPI interface.
 type MyGAPI struct {
 	Name     string // graph name
@@ -38,6 +50,39 @@ func NewMyGAPI(data gapi.Data, name string, interval uint) (*MyGAPI, error) {
 	return obj, obj.Init(data)
 }
 
+// Cli takes a cli.Context, and returns our GAPI if activated. All arguments
+// should take the prefix of the registered name. On activation, if there are
+// any validation problems, you should return an error. If this was not
+// activated, then you should return a nil GAPI and a nil error.
+func (obj *MyGAPI) Cli(c *cli.Context, fs resources.Fs) (*gapi.Deploy, error) {
+	if s := c.String(obj.Name); c.IsSet(obj.Name) {
+		if s != "" {
+			return nil, fmt.Errorf("input is not empty")
+		}
+
+		return &gapi.Deploy{
+			Name: obj.Name,
+			Noop: c.GlobalBool("noop"),
+			Sema: c.GlobalInt("sema"),
+			GAPI: &MyGAPI{
+			// TODO: add properties here...
+			},
+		}, nil
+	}
+	return nil, nil // we weren't activated!
+}
+
+// CliFlags returns a list of flags used by this deploy subcommand.
+func (obj *MyGAPI) CliFlags() []cli.Flag {
+	return []cli.Flag{
+		cli.StringFlag{
+			Name:  obj.Name,
+			Value: "",
+			Usage: "run",
+		},
+	}
+}
+
 // Init initializes the MyGAPI struct.
 func (obj *MyGAPI) Init(data gapi.Data) error {
 	if obj.initialized {
@@ -87,7 +132,7 @@ func (obj *MyGAPI) subGraph() (*pgraph.Graph, error) {
 // Graph returns a current Graph.
 func (obj *MyGAPI) Graph() (*pgraph.Graph, error) {
 	if !obj.initialized {
-		return nil, fmt.Errorf("libmgmt: MyGAPI is not initialized")
+		return nil, fmt.Errorf("%s: MyGAPI is not initialized", Name)
 	}
 
 	g, err := pgraph.NewGraph(obj.Name)
@@ -142,7 +187,7 @@ func (obj *MyGAPI) Next() chan gapi.Next {
 		defer close(ch) // this will run before the obj.wg.Done()
 		if !obj.initialized {
 			next := gapi.Next{
-				Err:  fmt.Errorf("libmgmt: MyGAPI is not initialized"),
+				Err:  fmt.Errorf("%s: MyGAPI is not initialized", Name),
 				Exit: true, // exit, b/c programming error?
 			}
 			ch <- next
@@ -171,7 +216,7 @@ func (obj *MyGAPI) Next() chan gapi.Next {
 				return
 			}
 
-			log.Printf("libmgmt: Generating new graph...")
+			log.Printf("%s: Generating new graph...", Name)
 			select {
 			case ch <- gapi.Next{}: // trigger a run
 			case <-obj.closeChan:
@@ -185,7 +230,7 @@ func (obj *MyGAPI) Next() chan gapi.Next {
 // Close shuts down the MyGAPI.
 func (obj *MyGAPI) Close() error {
 	if !obj.initialized {
-		return fmt.Errorf("libmgmt: MyGAPI is not initialized")
+		return fmt.Errorf("%s: MyGAPI is not initialized", Name)
 	}
 	close(obj.closeChan)
 	obj.wg.Wait()
@@ -197,19 +242,19 @@ func (obj *MyGAPI) Close() error {
 func Run() error {
 
 	obj := &mgmt.Main{}
-	obj.Program = "libmgmt" // TODO: set on compilation
-	obj.Version = "0.0.1"   // TODO: set on compilation
-	obj.TmpPrefix = true    // disable for easy debugging
+	obj.Program = Name    // TODO: set on compilation
+	obj.Version = "0.0.1" // TODO: set on compilation
+	obj.TmpPrefix = true  // disable for easy debugging
 	//prefix := "/tmp/testprefix/"
 	//obj.Prefix = &p // enable for easy debugging
 	obj.IdealClusterSize = -1
 	obj.ConvergedTimeout = -1
 	obj.Noop = false // FIXME: careful!
 
-	obj.GAPI = &MyGAPI{ // graph API
-		Name:     "libmgmt", // TODO: set on compilation
-		Interval: 60 * 10,   // arbitrarily change graph every 15 seconds
-	}
+	//obj.GAPI = &MyGAPI{ // graph API
+	//	Name:     Name, // TODO: set on compilation
+	//	Interval: 60 * 10,   // arbitrarily change graph every 15 seconds
+	//}
 
 	if err := obj.Init(); err != nil {
 		return err

examples/lib/libmgmt-subgraph1.go

@@ -14,6 +14,15 @@ import (
 	mgmt "github.com/purpleidea/mgmt/lib"
 	"github.com/purpleidea/mgmt/pgraph"
 	"github.com/purpleidea/mgmt/resources"
+
+	"github.com/urfave/cli"
+)
+
+// XXX: this has not been updated to latest GAPI/Deploy API. Patches welcome!
+
+const (
+	// Name is the name of this frontend.
+	Name = "libmgmt"
 )
 
 // MyGAPI implements the main GAPI interface.
@@ -36,6 +45,39 @@ func NewMyGAPI(data gapi.Data, name string, interval uint) (*MyGAPI, error) {
 	return obj, obj.Init(data)
 }
 
+// Cli takes a cli.Context, and returns our GAPI if activated. All arguments
+// should take the prefix of the registered name. On activation, if there are
+// any validation problems, you should return an error. If this was not
+// activated, then you should return a nil GAPI and a nil error.
+func (obj *MyGAPI) Cli(c *cli.Context, fs resources.Fs) (*gapi.Deploy, error) {
+	if s := c.String(obj.Name); c.IsSet(obj.Name) {
+		if s != "" {
+			return nil, fmt.Errorf("input is not empty")
+		}
+
+		return &gapi.Deploy{
+			Name: obj.Name,
+			Noop: c.GlobalBool("noop"),
+			Sema: c.GlobalInt("sema"),
+			GAPI: &MyGAPI{
+			// TODO: add properties here...
+			},
+		}, nil
+	}
+	return nil, nil // we weren't activated!
+}
+
+// CliFlags returns a list of flags used by this deploy subcommand.
+func (obj *MyGAPI) CliFlags() []cli.Flag {
+	return []cli.Flag{
+		cli.StringFlag{
+			Name:  obj.Name,
+			Value: "",
+			Usage: "run",
+		},
+	}
+}
+
 // Init initializes the MyGAPI struct.
 func (obj *MyGAPI) Init(data gapi.Data) error {
 	if obj.initialized {
@@ -53,7 +95,7 @@ func (obj *MyGAPI) Init(data gapi.Data) error {
 // Graph returns a current Graph.
 func (obj *MyGAPI) Graph() (*pgraph.Graph, error) {
 	if !obj.initialized {
-		return nil, fmt.Errorf("libmgmt: MyGAPI is not initialized")
+		return nil, fmt.Errorf("%s: MyGAPI is not initialized", Name)
 	}
 
 	g, err := pgraph.NewGraph(obj.Name)
@@ -132,7 +174,7 @@ func (obj *MyGAPI) Next() chan gapi.Next {
 		defer close(ch) // this will run before the obj.wg.Done()
 		if !obj.initialized {
 			next := gapi.Next{
-				Err:  fmt.Errorf("libmgmt: MyGAPI is not initialized"),
+				Err:  fmt.Errorf("%s: MyGAPI is not initialized", Name),
 				Exit: true, // exit, b/c programming error?
 			}
 			ch <- next
@@ -161,7 +203,7 @@ func (obj *MyGAPI) Next() chan gapi.Next {
 				return
 			}
 
-			log.Printf("libmgmt: Generating new graph...")
+			log.Printf("%s: Generating new graph...", Name)
 			select {
 			case ch <- gapi.Next{}: // trigger a run
 			case <-obj.closeChan:
@@ -175,7 +217,7 @@ func (obj *MyGAPI) Next() chan gapi.Next {
 // Close shuts down the MyGAPI.
 func (obj *MyGAPI) Close() error {
 	if !obj.initialized {
-		return fmt.Errorf("libmgmt: MyGAPI is not initialized")
+		return fmt.Errorf("%s: MyGAPI is not initialized", Name)
 	}
 	close(obj.closeChan)
 	obj.wg.Wait()
@@ -196,10 +238,10 @@ func Run() error {
 	obj.ConvergedTimeout = -1
 	obj.Noop = false // FIXME: careful!
 
-	obj.GAPI = &MyGAPI{ // graph API
-		Name:     "libmgmt", // TODO: set on compilation
-		Interval: 60 * 10,   // arbitrarily change graph every 15 seconds
-	}
+	//obj.GAPI = &MyGAPI{ // graph API
+	//	Name:     "libmgmt", // TODO: set on compilation
+	//	Interval: 60 * 10,   // arbitrarily change graph every 15 seconds
+	//}
 
 	if err := obj.Init(); err != nil {
 		return err

examples/lib/libmgmt1.go

@@ -15,6 +15,15 @@ import (
 	"github.com/purpleidea/mgmt/pgraph"
 	"github.com/purpleidea/mgmt/resources"
 	"github.com/purpleidea/mgmt/yamlgraph"
+
+	"github.com/urfave/cli"
+)
+
+// XXX: this has not been updated to latest GAPI/Deploy API. Patches welcome!
+
+const (
+	// Name is the name of this frontend.
+	Name = "libmgmt"
 )
 
 // MyGAPI implements the main GAPI interface.
@@ -37,6 +46,39 @@ func NewMyGAPI(data gapi.Data, name string, interval uint) (*MyGAPI, error) {
 	return obj, obj.Init(data)
 }
 
+// Cli takes a cli.Context, and returns our GAPI if activated. All arguments
+// should take the prefix of the registered name. On activation, if there are
+// any validation problems, you should return an error. If this was not
+// activated, then you should return a nil GAPI and a nil error.
+func (obj *MyGAPI) Cli(c *cli.Context, fs resources.Fs) (*gapi.Deploy, error) {
+	if s := c.String(obj.Name); c.IsSet(obj.Name) {
+		if s != "" {
+			return nil, fmt.Errorf("input is not empty")
+		}
+
+		return &gapi.Deploy{
+			Name: obj.Name,
+			Noop: c.GlobalBool("noop"),
+			Sema: c.GlobalInt("sema"),
+			GAPI: &MyGAPI{
+			// TODO: add properties here...
+			},
+		}, nil
+	}
+	return nil, nil // we weren't activated!
+}
+
+// CliFlags returns a list of flags used by this deploy subcommand.
+func (obj *MyGAPI) CliFlags() []cli.Flag {
+	return []cli.Flag{
+		cli.StringFlag{
+			Name:  obj.Name,
+			Value: "",
+			Usage: "run",
+		},
+	}
+}
+
 // Init initializes the MyGAPI struct.
 func (obj *MyGAPI) Init(data gapi.Data) error {
 	if obj.initialized {
@@ -54,7 +96,7 @@ func (obj *MyGAPI) Init(data gapi.Data) error {
 // Graph returns a current Graph.
 func (obj *MyGAPI) Graph() (*pgraph.Graph, error) {
 	if !obj.initialized {
-		return nil, fmt.Errorf("libmgmt: MyGAPI is not initialized")
+		return nil, fmt.Errorf("%s: MyGAPI is not initialized", Name)
 	}
 
 	n1, err := resources.NewNamedResource("noop", "noop1")
@@ -96,7 +138,7 @@ func (obj *MyGAPI) Next() chan gapi.Next {
 		defer close(ch) // this will run before the obj.wg.Done()
 		if !obj.initialized {
 			next := gapi.Next{
-				Err:  fmt.Errorf("libmgmt: MyGAPI is not initialized"),
+				Err:  fmt.Errorf("%s: MyGAPI is not initialized", Name),
 				Exit: true, // exit, b/c programming error?
 			}
 			ch <- next
@@ -124,7 +166,7 @@ func (obj *MyGAPI) Next() chan gapi.Next {
 				return
 			}
 
-			log.Printf("libmgmt: Generating new graph...")
+			log.Printf("%s: Generating new graph...", Name)
 			select {
 			case ch <- gapi.Next{}: // trigger a run
 			case <-obj.closeChan:
@@ -138,7 +180,7 @@ func (obj *MyGAPI) Next() chan gapi.Next {
 // Close shuts down the MyGAPI.
 func (obj *MyGAPI) Close() error {
 	if !obj.initialized {
-		return fmt.Errorf("libmgmt: MyGAPI is not initialized")
+		return fmt.Errorf("%s: MyGAPI is not initialized", Name)
 	}
 	close(obj.closeChan)
 	obj.wg.Wait()
@@ -157,10 +199,10 @@ func Run() error {
 	obj.ConvergedTimeout = -1
 	obj.Noop = true
 
-	obj.GAPI = &MyGAPI{ // graph API
-		Name:     "libmgmt", // TODO: set on compilation
-		Interval: 15,        // arbitrarily change graph every 15 seconds
-	}
+	//obj.GAPI = &MyGAPI{ // graph API
+	//	Name:     "libmgmt", // TODO: set on compilation
+	//	Interval: 15,        // arbitrarily change graph every 15 seconds
+	//}
 
 	if err := obj.Init(); err != nil {
 		return err

examples/lib/libmgmt2.go

@@ -15,6 +15,15 @@ import (
 	mgmt "github.com/purpleidea/mgmt/lib"
 	"github.com/purpleidea/mgmt/pgraph"
 	"github.com/purpleidea/mgmt/resources"
+
+	"github.com/urfave/cli"
+)
+
+// XXX: this has not been updated to latest GAPI/Deploy API. Patches welcome!
+
+const (
+	// Name is the name of this frontend.
+	Name = "libmgmt"
 )
 
 // MyGAPI implements the main GAPI interface.
@@ -39,6 +48,39 @@ func NewMyGAPI(data gapi.Data, name string, interval uint, count uint) (*MyGAPI,
 	return obj, obj.Init(data)
 }
 
+// Cli takes a cli.Context, and returns our GAPI if activated. All arguments
+// should take the prefix of the registered name. On activation, if there are
+// any validation problems, you should return an error. If this was not
+// activated, then you should return a nil GAPI and a nil error.
+func (obj *MyGAPI) Cli(c *cli.Context, fs resources.Fs) (*gapi.Deploy, error) {
+	if s := c.String(obj.Name); c.IsSet(obj.Name) {
+		if s != "" {
+			return nil, fmt.Errorf("input is not empty")
+		}
+
+		return &gapi.Deploy{
+			Name: obj.Name,
+			Noop: c.GlobalBool("noop"),
+			Sema: c.GlobalInt("sema"),
+			GAPI: &MyGAPI{
+			// TODO: add properties here...
+			},
+		}, nil
+	}
+	return nil, nil // we weren't activated!
+}
+
+// CliFlags returns a list of flags used by this deploy subcommand.
+func (obj *MyGAPI) CliFlags() []cli.Flag {
+	return []cli.Flag{
+		cli.StringFlag{
+			Name:  obj.Name,
+			Value: "",
+			Usage: "run",
+		},
+	}
+}
+
 // Init initializes the MyGAPI struct.
 func (obj *MyGAPI) Init(data gapi.Data) error {
 	if obj.initialized {
@@ -56,7 +98,7 @@ func (obj *MyGAPI) Init(data gapi.Data) error {
 // Graph returns a current Graph.
 func (obj *MyGAPI) Graph() (*pgraph.Graph, error) {
 	if !obj.initialized {
-		return nil, fmt.Errorf("libmgmt: MyGAPI is not initialized")
+		return nil, fmt.Errorf("%s: MyGAPI is not initialized", Name)
 	}
 
 	g, err := pgraph.NewGraph(obj.Name)
@@ -89,7 +131,7 @@ func (obj *MyGAPI) Next() chan gapi.Next {
 		defer close(ch) // this will run before the obj.wg.Done()
 		if !obj.initialized {
 			next := gapi.Next{
-				Err:  fmt.Errorf("libmgmt: MyGAPI is not initialized"),
+				Err:  fmt.Errorf("%s: MyGAPI is not initialized", Name),
 				Exit: true, // exit, b/c programming error?
 			}
 			ch <- next
@@ -117,7 +159,7 @@ func (obj *MyGAPI) Next() chan gapi.Next {
 				return
 			}
 
-			log.Printf("libmgmt: Generating new graph...")
+			log.Printf("%s: Generating new graph...", Name)
 			select {
 			case ch <- gapi.Next{}: // trigger a run
 			case <-obj.closeChan:
@@ -131,7 +173,7 @@ func (obj *MyGAPI) Next() chan gapi.Next {
 // Close shuts down the MyGAPI.
 func (obj *MyGAPI) Close() error {
 	if !obj.initialized {
-		return fmt.Errorf("libmgmt: MyGAPI is not initialized")
+		return fmt.Errorf("%s: MyGAPI is not initialized", Name)
 	}
 	close(obj.closeChan)
 	obj.wg.Wait()
@@ -150,11 +192,11 @@ func Run(count uint) error {
 	obj.ConvergedTimeout = -1
 	obj.Noop = true
 
-	obj.GAPI = &MyGAPI{ // graph API
-		Name:     "libmgmt", // TODO: set on compilation
-		Count:    count,     // number of vertices to add
-		Interval: 15,        // arbitrarily change graph every 15 seconds
-	}
+	//obj.GAPI = &MyGAPI{ // graph API
+	//	Name:     "libmgmt", // TODO: set on compilation
+	//	Count:    count,     // number of vertices to add
+	//	Interval: 15,        // arbitrarily change graph every 15 seconds
+	//}
 
 	if err := obj.Init(); err != nil {
 		return err