When the bazel remote cache was switched from buildbuddy to a self
hosted server in #8342, I asked that the buildbuddy hooks be remain so
that I could still use their caching service for local builds.
The downside of this was that my forks builds aren't leveraging
buildbuddy, so if I'm fiddling with something heavy like a wpimath
robotpy thing, my CI builds never update a cache and never are warm when
I push fixups.
This PR combines the `setup-bazel-remote` and `setup-build-buddy`
actions which set up the bazel remote cache. Rather than having two
different version, the correct one will be choosen in the following
order:
1. Use wpi's server with write access if the `bazel_remote` information
is set (This basically would only happen on main branch in
`wpilibsuite/allwpilib` since secrets aren't accessible from builds
originating in forks)
2. Use buildbuddy if the key it is present (This would work for my fork
builds)
3. Fall back to the readonly version of wpi's server
As seen
[here](https://github.com/bzlmodRio/allwpilib/actions/runs/25777428163/job/75712863120#step:7:28)
the build in my fork will run with buildbuddy, and my PR's build here
should fall back to readonly mode.
This is owned by WPI, so we can make it as big or small as we want.
This lets us check in more of the bazel build.
Signed-off-by: Austin Schuh <austin.linux@gmail.com>
The framework fundamentally relies on the continuation API added in Java 21 (which is currently internal to the JDK). Continuations allow for call stacks to be saved to the heap and resumed later.
The async framework allows command bodies to be written in an imperative style. However, an async command will need to be actively cooperative and periodically call coroutine.yield() in loops to yield control back to the command scheduler to let it process other commands.
There are also some other additions like priority levels (as opposed to a blanket yes/no for ignoring incoming commands), factories requiring names be provided for commands, and the scheduler tracking all running commands and not just the highest-level groups. However, those changes aren't unique to an async framework, and could just as easily be used in a traditional command framework.
This ensures that complete uniformity in how the generation scripts are run. All dependencies and scripts are set up in the exact same way, each time. The old pregen_all script has been removed and moved into the composite action to ensure failed scripts will always fail the job.
