00:00 DavidHeidelberg: I even took care about it in the script. it's effectively useless because first call is against the real repo, second it's cached... but I think I can do the removal of the cached on
00:04 DavidHeidelberg: ok, I found the problem git checkout "${BRANCH}" thou it's the "cached" origin, I need to specify the origin_gitlab rather
00:30 Company: my AMD claims it can't do RGB8 on Vulkan but it can do RGB8 on GLES - is that because Mesa has fallbacks for that due to the GLES spec demanding support?
00:30 Company: Or is that an oversight somewhere?
00:32 Company: I'm wondering because all image decoders (read: libpng and libjpeg) output RGB8 by default and I want to avoid having to convert too much
00:52 mareko: yes, fallbacks
00:52 mareko: GLES converts to RGBX on the CPU
01:00 Company: good to know
01:31 DavidHeidelberg: mareko: sent WIP MR, it should cover libdrm2 for alpine+debian, but I'll need to test it tomorrow
18:25 mareko: DavidHeidelberg: thanks, I don't have the expertise to review it though
18:57 DavidHeidelberg: np, I'm fighting a bit so far with s390x and ppc64le (no idea what can be different for these in terms of packaging from the original libdrm) and when I resolve this, we can bump everywhere