libglvnd abstracts away vendor-provided libGL. Currently, only Mesa and NVIDIA are supported. In future, multiple Mesa versions may co-exist. See review D25020 for the patch.
Can someone explain why exp-run hasn't been performed yet?
(In reply to Jan Beich from comment #1) According to the review the patch was not approved by maintainer.
(In reply to Antoine Brodin from comment #2) I'm disappointed in portmgr@ for facilitating the deadlock situation after x11@ said "it has to be thoroughly tested".
(In reply to Jan Beich from comment #3) Existing deadlock aside, this seems like an odd stance to take. If I'm maintainer of something that's worth doing an exp-run for, I likely will not approve of a patch until I also see the consequences of the patch on reverse dependencies. i.e. maybe I approve in principle, but I want to see that due diligence is done before putting that on paper. Antoine, do you have any numbers on how many exp-runs we end up not doing over some period of time because a maintainer doesn't approve of the patch? I think it makes sense to revise this, because I can't imagine this is saving us all that much, resource-wise.
(In reply to Kyle Evans from comment #4) I don't want to waste an exp-run if x11@ is going to reject the patch even if the exp-run is successfull. x11@ should speak up.
Ping. I can't improve the patch or fix consumers without an exp-run. x11@ actually did speak in favor years ago, see https://github.com/FreeBSDDesktop/freebsd-ports/commits/feature/libglvnd
(In reply to Jan Beich from comment #6) Can you refresh the patch?
New failure logs on 12.1 amd64: http://package18.nyi.freebsd.org/data/121amd64-default-foo/2021-01-31_19h12m55s/logs/errors/cubicsdr-0.2.5_5.log http://package18.nyi.freebsd.org/data/121amd64-default-foo/2021-01-31_19h12m55s/logs/errors/joequake-0.15b1329_9.log http://package18.nyi.freebsd.org/data/121amd64-default-foo/2021-01-31_19h12m55s/logs/errors/nehquake-1.0_8.log http://package18.nyi.freebsd.org/data/121amd64-default-foo/2021-01-31_19h12m55s/logs/errors/qudos-0.40.1_19.log http://package18.nyi.freebsd.org/data/121amd64-default-foo/2021-01-31_19h12m55s/logs/errors/tremulous-1.1.0_12.log http://package18.nyi.freebsd.org/data/121amd64-default-foo/2021-01-31_19h12m55s/logs/errors/kodi-19.0.r1_2.log http://package18.nyi.freebsd.org/data/121amd64-default-foo/2021-01-31_19h12m55s/logs/errors/muffin-4.8.0.log http://package18.nyi.freebsd.org/data/121amd64-default-foo/2021-01-31_19h12m55s/logs/errors/virglrenderer-0.8.2.log
(In reply to Antoine Brodin from comment #8) All fixed. Can you retry?
New failure logs on i386: http://pb2.nyi.freebsd.org/data/122i386-default-foo/2021-02-11_10h13m57s/logs/errors/q3cellshading-1.0_3.log http://pb2.nyi.freebsd.org/data/122i386-default-foo/2021-02-11_10h13m57s/logs/errors/quake3-1.32c_5.log On amd64 it looks fine
(In reply to Antoine Brodin from comment #10) Fixed. To find related commits see https://freshbsd.org/search?q=unbreak+libglvnd
(In reply to Antoine Brodin from comment #10) If exp-run is fine now, please, don't squat on Assignee field.