Resources that can be grouped into the http:server resource must have
that prefix. Grouping is basically hierarchical, and without that common
prefix, it means we'd have to special-case our grouping algorithm.
I inverted the logic for complex setups and forgot to handle the zero
cases. I also didn't notice my loop continue error. This cleans all this
up so that we can have proper exported resource matching.
Most of the time, we don't need to have a dynamic call sub graph, since
the actual function call could be represented statically as it
originally was before lambda functions were implemented. Simplifying the
graph shape has important performance benefits in terms of both keep the
graph smaller (memory, etc) and in avoiding the need to run transactions
at runtime (speed) to reshape the graph.
Co-authored-by: Samuel Gélineau <gelisam@gmail.com>
This commit implements a range function that mimicks python's range
built-in by having a start, stop, and range argument. There's also
a few examples and tests to mimick Python's examples to guarantee
we're consistent with their behaviour.
/ Test (basic) on ubuntu-latest with golang 1.23 (push) Has been cancelled
/ Test (race) on ubuntu-latest with golang 1.23 (push) Has been cancelled
/ Test (shell) on ubuntu-latest with golang 1.23 (push) Has been cancelled
This lets us look at the available resource data for collection, and to
filter it so we can decide what we want to collect on our machine.
Other types of collect functions could be added in the future.
I've been waiting to write this patch for a long time. I firmly believe
that the idea of "exported resources" was truly a brilliant one, but
which was never even properly understood by its original inventors! This
patch set aims to show how it should have been done.
The main differences are:
* Real-time modelling, since "once per run" makes no sense.
* Filter with code/functions not language syntax.
* Directed exporting to limit the intended recipients.
The next step is to add more "World" reading and filtering functions to
make it easy and expressive to make your selection of resources to
collect!
This is a basic implementation of a detection method for whether mgmt is
running in a virtualized environment. We achieve this by doing two types
of checks: on one hand, we check if the CPU flags can confirm the
presence of a virtualized env; on the other, we check if the presence
of known files related with DMI (and their contents) can confirm whether
we're inside a virt env. Either of these situations will cause the
function to return true, with the default case being false. All of these
checks are relatively naive and can be improved by looking at the main
inspiration for this implementation, which was systemd's own check for
the presence of virtualization.
This is a first attempt to add a new function for URL parsing, using
go's net/url package and the simple API. This is still a barebones
implementation, there's possibility to expose more information. It also
includes simple tests.
We forgot to reject this corner case which could lead to a runtime error
since the expected type from the incoming struct would not match what
we're handling.
Instead of constantly making these updates, let's just remove the year
since things are stored in git anyways, and this is not an actual modern
legal risk anymore.
This lets you boot from ipxe. You can run the ipxe shell from their
stock image or the netboot.xyz one. For the latter, press "m", then type
"dhcp" (machine is now pingable!) then type "route" to check the ip.
To boot type:
chain http://192.168.42.1:4280/menu.ipxe
and you're off!
Thanks to frebib for finding the workaround to the VFS bug. The answer
is you need to run the imgfree command to unblock the initrd.
Unfortunately with Fedora 41 and DNF5, this breaks what our host machine
expects. So you now need dnf5+ to make this work, or you can revert this
commit.
This is needed for bigger code bases. Remember: this is consulted at the
deploy stage, and the deploy contains the entire tree (including
modules) of everything it needs to run. This is why you don't need to
add a --module-path arg when running mgmt in systemd with the "empty"
frontend.
This adds a new util package with some useful functionality which could
be implemented as pure mcl, but instead we add it here as a good place
to help with code reuse.
With the release of Fedora 41, I was getting lots of mirror errors.
Hopefully this helps make it more robust. It was failing repeatedly
while trying to download packages, and I kept having to restart things,
but once I added this option things worked. Hopefully they're related.
