Closed BertalanD closed 1 month ago
I saw your patch on gcc-patches@ (thanks for doing this!), and the follow-on (which is correct). I suggest you repost with that fixed and cc the analyser maintainer (since most changes are there).
It would generally be preferable to pull in an upstream patch once it is applied; although I guess we probably will need to re-issue the darwin branches for the GCC-14, 13, 12 and 11 when that is done. If we do not get any traction with the patch in a reasonable timeframe - I will put it on the darwin branches here.
I'm currently fighting with git-send-email as I'm not too familiar with that workflow, but will try to post the updated patch soon.
I would like to have this backported for all active release branches upstream, too. I originally found this because SerenityOS contributors were reporting that they were failing to build our 13.2.0-based toolchain.
re-git send-email - you can attach a patch to the response to a review, if needed.
... we want to avoid too much code churn from an as yet unreleased OS and Xcode pair; tracking issues in betas is better done with Feedbacks to Apple (although if the response in this case is that this is not likely to be altered, we can proceed). Persevere with upstream - but I will apply here at the weekend to the Arm64 branch rebase - i'd really really prefer to wait for the upstream patch to back port because it means making another release (and my hardware is all tied up with the GCC-11.5 release for the next week) - but, again, if we get no upstream traction - then fine, we can do this.
OK .. so now it's approved upstream
if no to either - please add a signed-off-by line to your patch header and send me the git-format output directly - I can apply it for you.
I do not have commit access; sent you the patch directly via email with the DCO sign-off added.
Thanks for all the help!
this is fixed on trunk and I back ported to gcc-14 (for gcc-14.2) and 11.5 (darwin branch only) - I guess we could do an r2 for 13.3 and/or r1 for 12.4 if anyone has a pressing need - but 11.5 would provide a bootstrap compiler for D - so it does not seem too pressing to me.
Already posted a patch here: https://gcc.gnu.org/pipermail/gcc-patches/2024-July/656779.html
Looks like this needs to be solved on GCC's side; this response on the LLVM Discord indicates that the stddef.h change likely won't be reverted: