Commands v3 had a few changes due to the upgrade:
- Java 24 removed the Pinned: MONITOR IllegalStateException when
yielding in a synchronized block, so we no longer need to special case
for it
- Lambda method name generation was tweaked, requiring tests to be
updated
- Bazel java_rules needed to be bumped to support Java 25
Closes#8425
This primarily fixes up the bazel publishing to match the gradle
publishing again, as some new libraries were added but not hooked up to
the maven publishing.
During the process, I noticed that the 3rd party libraries (googletest,
catch2, and imgui_suite) were still getting published on the old
`edu.wpi` namespace. I tried to clean up all the other references to
that that I could. Note: opencv and libssh are handled outside
`allwpilib` so they need to be updated separately.
Jackson is a very heavy library; it supports loads of features that we
don't need, and historically has caused issues due to long class loading
times (a little over 2 seconds to load AprilTagFieldLayout). This often
manifests as a help request in the form of "my robot disables when I do
X, but doesn't disable when doing X in subsequent attempts until code
restart." While SC has brought down Jackson loading times significantly,
with AprilTagFieldLayout loads taking only 330 milliseconds, that's
still a rather long delay, and while libraries should handle any JSON
loading ahead of time to prevent delays in auto/teleop, it would still
be good to make the worst case better to reduce user frustration.
Benchmarks indicate using [Avaje
Jsonb](https://github.com/avaje/avaje-jsonb) to load AprilTagFieldLayout
only takes ~70 ms, a fair chunk of which isn't actually in Avaje Jsonb
(~4 ms is spent on using getResourceAsStream to retrieve the JSON file,
~8 ms is spent on just loading the AprilTag class and its dependencies).
Note that all times listed are end-to-end, meaning nothing else was done
except for the operation being benchmarked, and doing arithmetic on them
can be flawed due to some classes being loaded twice, i.e.,
getResourceAsStream and `new AprilTag()` likely load some of the same
JDK classes and so subtracting both from the Avaje Jsonb load time is
likely slightly incorrect because class loading is being double counted.
For our purposes, it's likely accurate enough and is mostly just for
contextualization.
Benchmarks were run on a Raspberry Pi CM5 with 2 GB of RAM. Source code
for the
[results](https://github.com/user-attachments/files/26471452/benchmark.txt)
can be found in the "Fastjson2" commit
(2456d15ca8ebd17635e607cd40bf8816e77869a1).
Avaje Jsonb uses code generation via annotation processors to generate
the classes needed to do JSON serde and uses service providers to find
them, which will require downstream changes in robot projects, as the
different service providers in each library must be merged together for
Avaje Jsonb to function. We will use the Gradle shadow plugin, as its
already used by the installer and therefore adds zero additional
dependencies.
`Color::FromHSV` didn't match the Java `Color.fromHSV` in some saturated
edge cases, introducing an off-by-one error when the HSV color should
correspond complete saturation of one or two of the primary colors.
Example:
- Java: `Color.fromHSV(0, 255, 255) -> (255, 0, 0)`
- C++: `Color::FromHSV(0, 255, 255) -> (255, 1, 1)`
This also means the following methods are also transitively affected:
- `AddressableLED::LEDData::SetHSV`
- `LEDPattern::Rainbow`
This off-by-one error is introduced by a rounding error from the chroma
calculation, which was dividing by 256 rather than the appropriate
maximum value of 255 like in Java:
7ca35e5678/wpilibj/src/main/java/edu/wpi/first/wpilibj/util/Color.java (L176-L177)
Also port appropriate tests from Java to C++ to catch this bug.
I found this bug when I tried to port `AddressableLEDBuffer` to RobotPy.
Codex found the root cause :)
Some discussion with the tech team showed that there were some real
advantages to being able to pass a 2nd type. It allows separating the DS
and Robot. Additionally, we can make the DriverStationBase class
actually usable instead of the existing DriverStation class which is
impossible to handle in intellisense because it has too much.
This won't fully be doable in C++, but we will need to implement
something similar in python.
This hooks up the bazel build to the robotpyExamples. It can use the
(formly pyfrc or whatever) automatic unit tests for an example, as well
as exposing the ability to run the example in simulation, with or
without `halsim_gui` with a command such as `bazel run
//robotpyExamples:AddressableLED-sim`
This required building and using wheels instead of just a normal
`py_library`, so that things like `ENTRY_POINTS` can be used. I took a
bare bones approach to building and naming the wheels (for example the
native ones don't have the OS info or python version in them, so they
wouldn't be suitable publish to pypi, but that can always be updated
later.
Also fixes the google compile-testing library to 0.23.0 (the latest
available at time of writing) instead of a wildcard
Jackson versions were inconsistent across projects; most were on 2.19.2,
but the fields subproject was on 2.15.2. All projects are now on 2.19.2
for consistency
Robotpy dropped support for 3.10 recently, so this updates to the next
lowest version. This will be helpful when the examples get merged into
`mostrobotpy`
There's not really a specific reason why I just jumped to 3.11 instead
of a newer version. If anybody suggests otherwise I will gladly bump it
higher.
Semiwrap / meson / robotpy define `NDEBUG` when building their software
in all modes, while `allwplib` only does it when building debug. This
causes the size of `DenseMap` to differ between the shared libraries
built here, and the extension modules built in `mostrobotpy`, causing
segfaults when you try to execute code that uses `DenseMap`. This is not
a problem with the robotpy code in `allwpilib`, because bazel uses the
exact same compiler flags when building the shared libraries and
pybind11 extensions.
Resync with `mostrobotpy`
This mostly involves the big "ignore almost everything in the HAL
project" and some fixups for the Addressable LED classes.
Required two small hand fixes to get it building over here with bazel,
and with more compiler warnings on.
I also manually zeroed out the `repo_url` field in the toml files to
avoid unnecessary churn whenever it goes from a release build to a
development build. I already did this with `version` field in there, and
will do a follow up PR that updates the copybara script to do it
automatically.
---------
Co-authored-by: Default email <default@default.com>
This adds a type caster for `WPI_String` so that pybind11 can more
easily auto-convert between that and strings. This helps remove the need
to do things like
[this](f1d77244c3)
in the opmodes fixup
This got missed in the reorg, and these values aren't actually used for
anything when building in `allwpilb`, but we might as well fix them here
to make the copybara process easier.
After replacing the remaining include guards with `#pragma once`, I was
able to merge all the wpiformat configs into one file in the repo root.
This should make the config easier to reason about and maintain in the
future.
