This adds a new test for readfile. Interestingly, it actually caught a
small bug, which was also fixed with this commit. I think the bug was
actually always masked, because it only occurred on shutdown, and in
this case we often don't care about how the stream exited, but it's a
good example of how a test case focused on just one small aspect can be
important.
As an aside, this test case also would have caught the bug fixed in
94c40909cc and by reverting that patch it
indeed fails.
This adds the ability to test that functions return the expected
streams, and to model this behaviour over time. This is done via a
"timeline" which runs an ordered list of actions that can both push new
values into the function stream, and wait and expect particular outputs.
Hopefully this will make our function implementations more robust!
This adds a function runner that runs pure functions. It will hopefully
be useful for speculative execution of functions for compile time
determination of types.
Readfile had a bug where it sent an empty string on startup. This has
ben fixed, and it now waits until the file contents are ready before
sending a string.
This unbreaks the mcl bindata code. Of course we could change the parser
to allow this prefix, but this is cleaner. The packages still have a
core prefix, which it seems we could also remove, but this isn't
particularly important for anything.
Due to a limitation in the template library, we need to rename some
functions. It's probably worth looking into modifying this library or
finding an alternate version.
After some investigation, it appears that SocketSet.Shutdown() and
SocketSet.Close() are not synchronous operations. The sendto system call
called in SocketSet.Shutdown() is not a blocking send. That means there
is a race in which SocketSet.Shutdown() sends a message to a file
descriptor to unblock select, while SocketSet.Close() will close the
file descriptor that the message is being sent to. If SocketSet.Close()
wins the race, select is listening on a dead file descriptor and will
hang indefinitely.
This is fixed in the current master by putting SocketSet.Close() inside
of the goroutine in which data from the socket is being received. It
relies on SocketSet.Shutdown() being called to terminate the goroutine.
While this works most of the time, there is a race here. All the
goroutines can also be terminated by a closeChan. If the goroutine
receives an event (thus unblocking select) and then closeChan is
triggered, both SocketSet.Shutdown() and SocketSet.Close() race, leading
to undefined behavior.
This patch ensures the ordering of the two function calls by pulling
them both out of the goroutine and separating them with a WaitGroup.
Co-authored-by: James Shubin <james@shubin.ca>
Adds a CPU count fact, that can be used to determine how many CPUs are
presently on the machine and ready for use (online). We get this by
reading from a netlink socket to the kernel, and the kernel sends us
uevents when CPUs are added, removed, and brought online or offline.
Whenever one of these events are received, we look in sysfs to update
the fact's Stream with the number of online CPUs.
This enables imports in mcl code, and is one of last remaining blockers
to using mgmt. Now we can start writing standalone modules, and adding
standard library functions as needed. There's still lots to do, but this
was a big missing piece. It was much harder to get right than I had
expected, but I think it's solid!
This unfortunately large commit is the result of some wild hacking I've
been doing for the past little while. It's the result of a rebase that
broke many "wip" commits that tracked my private progress, into
something that's not gratuitously messy for our git logs. Since this was
a learning and discovery process for me, I've "erased" the confusing git
history that wouldn't have helped. I'm happy to discuss the dead-ends,
and a small portion of that code was even left in for possible future
use.
This patch includes:
* A change to the cli interface:
You now specify the front-end explicitly, instead of leaving it up to
the front-end to decide when to "activate". For example, instead of:
mgmt run --lang code.mcl
we now do:
mgmt run lang --lang code.mcl
We might rename the --lang flag in the future to avoid the awkward word
repetition. Suggestions welcome, but I'm considering "input". One
side-effect of this change, is that flags which are "engine" specific
now must be specified with "run" before the front-end name. Eg:
mgmt run --tmp-prefix lang --lang code.mcl
instead of putting --tmp-prefix at the end. We also changed the GAPI
slightly, but I've patched all code that used it. This also makes things
consistent with the "deploy" command.
* The deploys are more robust and let you deploy after a run
This has been vastly improved and let's mgmt really run as a smart
engine that can handle different workloads. If you don't want to deploy
when you've started with `run` or if one comes in, you can use the
--no-watch-deploy option to block new deploys.
* The import statement exists and works!
We now have a working `import` statement. Read the docs, and try it out.
I think it's quite elegant how it fits in with `SetScope`. Have a look.
As a result, we now have some built-in functions available in modules.
This also adds the metadata.yaml entry-point for all modules. Have a
look at the examples or the tests. The bulk of the patch is to support
this.
* Improved lang input parsing code:
I re-wrote the parsing that determined what ran when we passed different
things to --lang. Deciding between running an mcl file or raw code is
now handled in a more intelligent, and re-usable way. See the inputs.go
file if you want to have a look. One casualty is that you can't stream
code from stdin *directly* to the front-end, it's encapsulated into a
deploy first. You can still use stdin though! I doubt anyone will notice
this change.
* The scope was extended to include functions and classes:
Go forth and import lovely code. All these exist in scopes now, and can
be re-used!
* Function calls actually use the scope now. Glad I got this sorted out.
* There is import cycle detection for modules!
Yes, this is another dag. I think that's #4. I guess they're useful.
* A ton of tests and new test infra was added!
This should make it much easier to add new tests that run mcl code. Have
a look at TestAstFunc1 to see how to add more of these.
As usual, I'll try to keep these commits smaller in the future!
This should prepare us so that we can build native mcl code alongside
the core *.go files which we already have. This includes a single mcl
file that is used as a placeholder so that the build doesn't fail if we
don't have any mcl files in the core/ directory. It will get ignored
automatically.
This is a giant refactor to move functions into a hierarchial module
layout. While this isn't entirely implemented yet, it should work
correctly once all the import bits have landed. What's broken at the
moment is the template function, which currently doesn't understand the
period separator.
This requires breaking changes in gofmt. It is hilarious that this was
changed. Oh well. This also moves to the latest stable etcd. Lastly,
this changes the `go vet` testing to test by package, since the new go
vet changed how it works and now fails without this change.
It turns out that some planned additions to the parser make it so that
the map type definition can be ambiguous. As a result, this patch
updates the definition so that the map definition is not confused with
an open curly bracket anywhere.
Thanks to pestle and stbenjamin for their help understanding yacc!
This allows golang tests to be marked as root or !root using build tags.
The matching tests are then run as expected using our test runner.
This also disables test caching which is unfriendly to repeated test
running and is an absurd golang default to add.
Lastly this hooks up the testing verbose flag to tests that accept a
debug variable.
These tests aren't enabled on travis yet because of how it installs
golang.
This giant patch makes some much needed improvements to the code base.
* The engine has been rewritten and lives within engine/graph/
* All of the common interfaces and code now live in engine/
* All of the resources are in one package called engine/resources/
* The Res API can use different "traits" from engine/traits/
* The Res API has been simplified to hide many of the old internals
* The Watch & Process loops were previously inverted, but is now fixed
* The likelihood of package cycles has been reduced drastically
* And much, much more...
Unfortunately, some code had to be temporarily removed. The remote code
had to be taken out, as did the prometheus code. We hope to have these
back in new forms as soon as possible.
The golang race detector complains about some unimportant races, and as
a result, this patch adds some mutexes to prevent these test failures.
We actually lock more than necessary, because a more accurate version
would be more time consuming to implement. Secondarily, it's likely that
in the future we replace this function graph algorithm with something
that is guaranteed to be glitch-free and supports back pressure.
I noticed a very intermittent test failure where interpret would end up
running, but *fail* because a value wasn't present. This should never
happen, because the function engine is designed to only call interpret
when there has been at least one value produced for every node in the
AST. So what is the bug that would produce:
interpret error: could not interpret: func value does not yet exist
About 20 minutes ago while I was getting to bed, it occurred to me where
to look! Out of bed and to the laptop, and after briefly reminding
myself of the code, I think I've found the issue.
What I think was happening, was that an AST node would produce a value,
and send a message on the aggregate channel. This channel is monitored,
and every time it receives a message, it checks to ensure that all the
values now exist before producing a message for interpret to run.
However, this AST node was not the final one to be produced, but before
the message was read by the aggregate channel, the last remaining AST
node ran and set it's "loaded" state to `true`, but *before* its value
was made available for the aggregate channel to read. That channel then
occasionally won the race and tried to access a value before it existed,
thus causing out intermittent bug.
At least I think that's what was going on. Hopefully this patch fixes
this, if not, then there's another bug hiding too! And of course, this
entire function engine could do with some proper analysis from someone
familiar with glitches, back pressure, and FRP parallelism.
One particular note was that I used my brain, not some fancy debugging
tool to find this. Maybe skilled debuggers can fork lift their tools
onto this type of problem, but I haven't those skills!
¯\_(ツ)_/¯
This adds a simple API for adding static, polymorphic, pure functions.
This lets you define a list of type signatures and the associated
implementations to overload a particular function name. The internals of
this API then do all of the hard work of matching the available
signatures to what statically type checks, and then calling the
appropriate implementation.
While this seems as if this would only work for function polymorphism
with a finite number of possible types, while this is mostly true, it
also allows you to add the `variant` "wildcard" type into your
signatures which will allow you to match a wider set of signatures.
A canonical use case for this is the len function which can determine
the length of both lists and maps with any contained type. (Either the
type of the list elements, or the types of the map keys and values.)
When using this functionality, you must be careful to ensure that there
is only a single mapping from possible type to signature so that the
"dynamic dispatch" of the function is unique.
It is worth noting that this API won't cover functions which support an
arbitrary number of input arguments. The well-known case of this,
printf, is implemented with the more general function API which is more
complicated.
This patch also adds some necessary library improvements for comparing
types to partial types, and to types containing variants.
Lastly, this fixes a bug in the `NewType` parser which parsed certain
complex function types wrong.
I forgot to handle the special case of a function using this API that
received no inputs. It was waiting for the first input to come in, and
as a result was never producing any output.
Remember that functions like this should *almost* be thought of as
constants of the system. You would expect their output to never change
during the lifetime of a particular program invocation.
This patch adds a simple function API for writing simple, pure
functions. This should reduce the amount of boilerplate required for
most functions, and make growing a stdlib significantly easier. If you
need to build more complex, event-generating functions, or statically
polymorphic functions, then you'll still need to use the normal API for
now.
This also makes all of these pure functions available automatically
within templates. It might make sense to group these functions into
packages to make their logical organization easier, but this is a good
enough start for now.
Lastly, this added some missing pieces to our types library. You can now
use `ValueOf` to convert from a `reflect.Value` to the corresponding
`Value` in our type system, if an equivalent exists.
Unfortunately, we're severely lacking in tests for these new types
library additions, but look forward to growing some in the future!
The test for gometalinter got silently broken in an earlier commit.
Look for the missing space that was added back in this commit to see
why! In any case, this now fixes some of the things that weren't
previously caught by this change.
If anyone knows how to run these sorts of tests properly so that entire
packages are tested and so that we can enable additional tests, please
let me know!
It's also unclear why goreportcard catches a few additional problems
which aren't found by running this ourselves.
See:
https://goreportcard.com/report/github.com/purpleidea/mgmt
for more information.
This is an initial implementation of the mgmt language. It is a
declarative (immutable) functional, reactive, domain specific
programming language. It is intended to be a language that is:
* safe
* powerful
* easy to reason about
With these properties, we hope this language, and the mgmt engine will
allow you to model the real-time systems that you'd like to automate.
This also includes a number of other associated changes. Sorry for the
large size of this patch.
