On Wed, 04 Sep 2024 at 20:38:04 +0200, John Paul Adrian Glaubitz wrote:
> On Wed, 2024-09-04 at 18:51 +0100, Simon McVittie wrote:
> > but it's taking a while (mips64el machines are not fast)
>
> The powerpc/ppc64 porterbox perotto.debian.net should be considerably faster
> than any mips64el machine we have. So, you may give it a try there as well.
Thanks for the suggestion, but gtk4's build-deps are unsatisfiable on
-ports, because freerdp2 can't be rebuilt for the new ffmpeg until its
FTBFS bugs are fixed, the current freerdp2 can't be installed because -ports
doesn't keep "cruft" packages like the old ffmpeg, and it's a dependency
for Weston.
So the best I'd be able to do is running less than half of the test suite:
some tests are run twice (under Xvfb and Weston) and some are only run
under Weston. Not a great way to have confidence that everything works!
The good news is that the removal of softpipe has been reverted in Mesa,
so we can continue to use gtk4's GALLIUM_DRIVER=softpipe workaround
on mips*, powerpc and sparc*, and probably add it to riscv* too. The
same workaround could probably help some other -ports architectures
(especially loong64) if someone wants to try it - although that isn't
actionable right now, again because of Weston.
smcv
