Change TF2Error names to be a bit more descriptive.

[clalancette]

The driving reason behind this is to avoid the NO_ERROR macro
on Windows. We can't totally get around the problem right now,
but we can deprecate the old names and switch to new ones.
After Galactic, we can remove the old names and get rid of the
problem completely.

Signed-off-by: Chris Lalancette [email protected]

This obsoletes #247 (which I'll go ahead and close out).

[wjwwood]

You could have a message with the deprecation, like "use TF2_X instead", but otherwise lgtm.

[clalancette]

You could have a message with the deprecation, like "use TF2_X instead", but otherwise lgtm.

Yeah, good point. I added that in d711136 . I'll run CI now:

• Linux
• Linux-aarch64
• macOS
• Windows

[clalancette]

The Windows warnings are known, and are being addressed elsewhere.

The macOS warnings are annoying. They are saying that [[deprecation]] on an enum value is a C++17-only feature, and since everything here is currently C++-14, we get the warning.

One way out of this conundrum is to wait an see what happens with ros-infrastructure/rep#291 . If we decide to switch to C++17, then we can just do that and use the deprecation as-is. The other thing we can possibly do is to deprecate the enum entirely, and come up with a new enum with new names. I'm going to hold off on this one for now.

[wjwwood]

on an enum value is a C++17-only feature, and since everything here is currently C++-14, we get the warning.

Oh yeah, that came up in the past I think.

[clalancette]

All right, since we can now switch to C++17, I've done that in this PR. Let's see what CI looks like now:

• Linux
• Linux-aarch64
• macOS
• Windows

[clalancette]

And of course now Windows is annoyed at C++17 changes. Sigh. OK, this is (still) a puzzle for another day.

[clalancette]

Given changes elsewhere, here is another run on Windows to see if it is any happier:

• Windows

[ahcorde]

LGTM, this may generate some deprecation warnings on some other packages which it would be great to track. I think we should run CI with --packages-above-and-dependencies tf2

[clalancette]

Instead of using C++17, which causes some downstream breakage, I think the simpler thing to do here is to ignore the warning that clang throws on macOS. This whole thing will eventually go away (for I-Turtle) anyway, so I think that is a reasonable enough thing to do. The latest commit does exactly that.

LGTM, this may generate some deprecation warnings on some other packages which it would be great to track.

My earlier CI attempts show that it doesn't, at least not in the core. There may be some breakage outside of the core, but that's why it is deprecated, not removed :).

run CI with --packages-above-and-dependencies tf2

Yep, definitely. Will run new CI with the latest commit and everything above tf2.

[clalancette]

CI:

• Linux
• Linux-aarch64
• macOS
• Windows

[clalancette]

CI is green here, so I'm going to go ahead and merge this. Thanks for the reviews.

@ahcorde If you have time, could you please review the release note I added for this at ros2/ros2_documentation#1957 ? Thanks.