Adds methods to compute the dot and cross products between Translation2ds and Translation3ds, as well as methods to compute the square of Distance and Norm, which allows avoiding some calls to sqrt in many cases.
Co-authored-by: Tyler Veness <calcmogul@gmail.com>
Since they are in different directories, they need to be special cased
This is needed to support wpilibsuite/WPILibInstaller-Avalonia#492,
since those are currently handled by special scripts.
The generator was calling a binary of the wrong architecture and failing
internally, but passing the build (incorrectly). Explode right away.
Signed-off-by: Austin Schuh <austin.linux@gmail.com>
It's only syntactic sugar over the CommandScheduler's schedule method and creates a footgun because it’s too obvious to try to use in incorrect places.
Co-authored-by: Starlight220 <53231611+Starlight220@users.noreply.github.com>
* `-Wfixed-enum-extension` was replaced with `-Wc23-extensions`
* Removed unused private variables in SysId
* Suppressed `-Wnontrivial-memcall` in imgui.h and imgui_internal.h
Ensures java deprecated notation is paired with javadoc and vice versa.
Adds javadoc deprecation for MecanumControllerCommand, ArmFeedForward,
ElevatorFeedforward, and MecanumDriveMotorVoltages
Fixes#7736
Supersedes #7737
Co-authored-by: Joseph Eng <91924258+KangarooKoala@users.noreply.github.com>
A noteworthy change is the replacement of the `dp.startswith(os.path.join(".", "subdir"))` pattern. pathlib doesn't offer something with similar semantics besides `match` and `full_match`, so there's now a helper function that replicates the behavior.
Other notable changes include the addition of type annotations to ensure code correctness, using == to check file names instead of `endswith` for clarity (`endswith` is still used to check extensions), manual walking and copying being refactored in googletest, json, memory, nanopb, protobuf, and sleipnir to use `walk_cwd_and_copy_if`, and matching functions being shortened to the point where they can just be inlined into the lambda.
Co-authored-by: Tyler Veness <calcmogul@gmail.com>
Co-authored-by: David Vo <auscompgeek@users.noreply.github.com>
These were apparently a meme about state being hard to manage rather
than a statement about the code itself. I spent a while trying to find
some complex logic this comment was alluding to that would indicate why
it's "a nightmare to manage".
This reverts commit 3dee19a435.
This was merged without sufficient review or discussion as to whether these units are value-add for the Java units library.
Addresses an issue where certain USB cameras, specifically the ArduCam OV9281, would freeze when attempting to stream on macOS.
The previous logic started the AVCaptureSession (startRunning) before locking the device for configuration (lockForConfiguration). While this works for many cameras, it causes the OV9281 to become unresponsive.
Further investigation revealed:
- Moving startRunning to after unlockForConfiguration resulted in macOS overriding the custom format and frame rate settings applied within the lock.
- The reliable solution, inspired by findings shared in the community (e.g., Stack Overflow), is to lock the device, apply the configuration, start the session, and then unlock the device.
This commit reorders the operations within deviceStreamOn in UsbCameraImplObjc.mm to follow the sequence: lockForConfiguration -> apply settings -> startRunning -> unlockForConfiguration. This ensures the desired camera configuration is applied correctly without causing device freezes on problematic hardware like the OV9281.
