Please check if your packages build (correctly) with GCC 14
Hello, we would like to ask Factory package maintainers to help us to bring GCC 14 to Tumbleweed. Currently there are 766 packages that are known to fail to build with it in one way or another. If you want to see if your package is one of them, have a look at: https://build.opensuse.org/project/monitor/home:rguenther:plgrnd?arch_x86_64=1&defaults=0&failed=1&repo_openSUSE_Factory=1 In order to reproduce the build using GCC 14 locally, you can invoke osc with: osc -A https://api.opensuse.org build --alternative-project home:rguenther:nextgcc Most of the failures are going to be because GCC 14 by default treats as errors a number of situations which up to now only triggered warnings, such as implicit int types or implicit function declarations. There is a very nice document describing most likely way to fix these and other issues along with possible pitfalls at: https://gcc.gnu.org/gcc-14/porting_to.html Please note that the new errors can and unfortunately also do affect environment probing, especially with old autotools. So even when your package builds, it is advisable to check and/or compare config.log and/or config.h with compilation using GCC 13. It is always best to fix the new errors, but if your project is just written in pre-C99-standard C, you can simply use option -std=gnu89 to revert to previous behavior in this regard. In the unfortunate case that the C used is more "modern" but relies on the use of this newly erroneous behavior and there simply isn't enough time to fix it soon, you can turn the errors back to warnings with a new C option -fpermissive. In that case, please report a bug and make it depend on meta-bug boo#1220571. If you want to have a look whether and/or how people at Gentoo and Fedore already dealt with such issues in their corresponding package, you can check out: - https://bugs.gentoo.org/showdependencytree.cgi?id=870412&hide_resolved=0 - https://gitlab.com/fweimer-rh/fedora-modernc/ Of course, there may be also other issues with the new compiler, please use meta-bug boo#1220574 to track these. If you suspect the new compiler is to blame, please (double check and then) indicate so in the bug and we will try to help you. Thank you very much, Martin
Tue, 30 Apr 2024 23:34:02 +0200 Martin Jambor <mjambor@suse.cz>:
These results are inaccurate, likely due to rebuild==local. For example, pkg "ocaml" builds, and has a certain %version-%release. The pkg "brltty" uses this exact %version-%release according to the logs. But the error message indicates that it uses another compiler. Apparently "brltty" started to build before "ocaml" was finished. Not a big deal. Olaf
On Fri, 3 May 2024, Olaf Hering wrote:
Tue, 30 Apr 2024 23:34:02 +0200 Martin Jambor <mjambor@suse.cz>:
These results are inaccurate, likely due to rebuild==local. For example, pkg "ocaml" builds, and has a certain %version-%release. The pkg "brltty" uses this exact %version-%release according to the logs. But the error message indicates that it uses another compiler. Apparently "brltty" started to build before "ocaml" was finished.
Note the project has useforbuild off, but source changes after the original build will now possibly run into ordering issues since the project builds against 'standard' and not 'snapshot' (that had other issues).
Not a big deal.
Yeah, it should only give a first idea of the number of issues but most importantly include all Factory packages, not just those tested by the stagings. Richard. -- Richard Biener <rguenther@suse.de> SUSE Software Solutions Germany GmbH, Frankenstrasse 146, 90461 Nuernberg, Germany; GF: Ivo Totev, Andrew McDonald, Werner Knoblich; (HRB 36809, AG Nuernberg)
participants (3)
-
Martin Jambor
-
Olaf Hering
-
Richard Biener