Please make a gcc 10.1.0 compatible release or pre-release available

[Flexximilian]

Since gcc 10.1.0 was officially released on May, 7th 2020, it would be truly helpful to have an official (pre-)release tag that builds on it. #503 and #509 were fixed, but a pre-release will be needed to get at least ~amd64 (unstable) support on Gentoo, and likely other distributions.

Thank you for your great work!

[uartie]

@xhaihao can you do a maintenance release?

[Flexximilian]

Just so I don't have to ask you twice if any other error would pop up, I've tried merging a live ebuild of commit 625d265 (current master/HEAD) on my up-to-date ~amd64 Gentoo with gcc 10.1.0, and it built just fine. That's great news for a quick adoption in the Gentoo main repo, and that there aren't any more gcc 10.1.0 issues (using a fresh toolchain) as far as I can tell.

Once you have a release out, I'll test it once more and issue an ebuild version bump request to the Gentoo Bugzilla.

If there's anything more I can help with, let me know.

[xhaihao]

@uartie @Flexximilian Thanks for reminding me. We will cut a new release soon.

[rubyFeedback]

GCC 10.1.0 causes quite some issues in general; not just here but elsewhere too, not only in gentoo. Not really having much to contribute here :) but I just wanted to mention it; even the ruby core team had to report a change upstream. Gues they changed a lot from 9 -> 10 in GCC.

[Flexximilian]

No, not really. One flag's default was flipped (not by accident and after a long discussion about it). Gcc 10 sets -fno-common by default, and that made very long standing bugs (which were handled gracefully due to gcc-9's default of -fcommon) now cause compile errors. In other words: gcc got stricter by default. Now you can explicitly set -fcommon to get the old behaviour, or fix the bugs (multiple definitions of the same identifier, basically), which is obviously the preferred way of dealing with this. A lot of packages actually have been fixed already in their code base, but often no new release containing the fix has been made / marked stable yet, but that's typically not because of any of the GCC 10 related corrections, but other, unrelated changes they happen to coincide with.

[xhaihao]

new version of intel-vaapi-driver: https://github.com/intel/intel-vaapi-driver/releases/tag/2.4.1