#11910 Mass Rebuild Fedora 40 problem
Closed: Invalid by smooge. Opened by smooge.

  • Describe the issue

yselkowitz
ugh, could we just not do -Werror=incompatible-pointer-types on i686? a lot of F40FTBFS are i686-only with such errors (e.g. uint64_t (aka long long unsigned int ) doesn't match long unsigned int *, etc.)

Is it possible for i686 to not have this as an error?


Maybe I'm just wishing there weren't so many new build failures with GCC 14 and the Modern C changes, but given the limited use of i686 packages, I think there should be some consideration of the cost/benefit of "fixing" code just for 32-bit compatibility.

A few examples:

gnome-keyring: https://koji.fedoraproject.org/koji/taskinfo?taskID=112010040
ldns: https://koji.fedoraproject.org/koji/taskinfo?taskID=112074780
ledmon: https://koji.fedoraproject.org/koji/taskinfo?taskID=112074925
libfabric: https://koji.fedoraproject.org/koji/taskinfo?taskID=112077476
libgphoto2: https://koji.fedoraproject.org/koji/taskinfo?taskID=112078167

While those that are leaf packages can be handled through https://fedoraproject.org/wiki/Changes/EncourageI686LeafRemoval , that still requires per-package work, and some clearly do not fall in that category.

Metadata Update from @phsmoura:
- Issue tagged with: medium-gain, medium-trouble, ops

So... I think this needs discussion with tools folks and fesco... if you are saying we should stop the mass rebuild again for this?

This devel thread: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/thread/GKCRXZESHSCCEKED2N5GNQ7GH32VSK2X/#F5WWUBXOFQDPW5NMH2YCC45LLFLSA32T
discusses it.

So, I would say starting there might be be good to see if tools folks even are willing to change anything here?

Was NOT suggesting a stop, there is no miscompilation here (afaik), just a question if it's worth the extra strictness for i686 given all the i686-only failures it is causing. If we do anything here, I think it could just be tweaking the defaults flags for i686 and retrying the relevant failed builds.

If this isn't the place to discuss this, feel free to close.

Hey sorry if this was the wrong place. I saw this in devel yesterday and wasn't sure what effect it would have on mass rebuild since a failure on x86_32 would kill the x86_64, etc build. With everyone focused on ipa I figured getting it captured for releng was good, but I guess this is a FESCO item.

Hello @smooge, @yselkowitz

We have one more issue opened here for tracking mass rebuild discussions if you want to use that #11888, or if you want any action items from here please let us know, and I can bring it us to our releng call otherwise we can close this one!

No I think this is a closed issue now.

Metadata Update from @smooge:
- Issue close_status updated to: Invalid
- Issue status updated to: Closed (was: Open)

Metadata