an explicit platform split matching the helper header's own condition.

[fujitatomoya]

Description

replaces #586

note: I first applied ATOMIC_VAR_INIT(-1) as suggested, but when I tested it in C23 mode (gcc -std=gnu2x), it failed — ATOMIC_VAR_INIT is already removed from this toolchain's <stdatomic.h>, producing "implicit declaration of function 'ATOMIC_VAR_INIT'" followed by "initializer element is not constant". which isn't theoretical. I switched to the fallback to have an explicit platform split matching the helper header's own condition.

Fixes # (issue)

Is this user-facing behavior change?

No

Did you use Generative AI?

No

Additional Information

[mergify]

Tick the box to add this pull request to the merge queue (same as @mergifyio queue).

• Queue this pull request

[fujitatomoya]

Pulls: #587
Gist: https://gist.githubusercontent.com/fujitatomoya/c7dc0267a7c77d0fcc0a827d492d036a/raw/cc34078e8e7c2d098313e1b7058f0062ee6c68ab/ros2.repos
BUILD args: --packages-above-and-dependencies rcutils
TEST args: --packages-above rcutils
ROS Distro: rolling
Job: ci_launcher
ci_launcher ran: https://ci.ros2.org/job/ci_launcher/20102

• Linux
• Linux-aarch64
• Linux-rhel
• Windows

[fujitatomoya]

i think this is a best option we got, any thoughs or better idea?

[yukaitung]

Without PR

With PR

The package builds successfully, lgtm, please backport to lyrical

This fix should be the best option since there is no way to ask Apple Clang to bypass it.

[fujitatomoya]

• Linux
• Linux-aarch64
• Linux-rhel
• Windows

[wentasah]

This works for me. Thanks!

[fujitatomoya]

• Windows

[ahcorde]

@fujitatomoya should we backport this one ?