00:28DUOLabs[m]: In venus, after CreateInstance is called, what happens next? Is control sent back to the calling application, as I don't see references in the source code to any further logic, just a definition of vn_CreateInstance.
03:04Jesica: Come Visit us @ irc.twistednet.org channel #twisted
10:13deathmist1: hey, about to bisect artifacting issues on Mesa 23.1.1 which weren't there on 23.0.3, but with LD_LIBRARY_PATH="builddir/src/glx" LIBGL_DRIVERS_PATH="builddir/src/gallium/targets/dri" running something gives libEGL fatal: DRI driver not from this Mesa build ('23.2.0-devel (git-92a7cba4f2)' vs '23.0.3')
10:14deathmist1: 23.0.3 is my system-wide package installation currently and I'd rather run the builds from the build tree instead of building new packages each time
10:15deathmist1: gpu is RX 6600 fwiw, perhaps people here already know about some rather intense artifacting especially seen on GNOME which makes extensive use of GPU accelerated GTK4 stuff
10:23eric_engestrom: deathmist1: the error means your loader and driver are not both from the same build; try using `meson devenv` to override everything together (https://mesonbuild.com/Commands.html#devenv) instead of setting env vars yourself
10:24deathmist1: cheers, I didn't even know that was a meson feature :)
11:05deathmist1: hm on that shell it instead "libEGL warning: MESA-LOADER: failed to open radeonsi: Error loading shared library /home/deathmist/mesa/builddir/src/gallium/targets/dri/radeonsi_dri.so: No such file or directory (search paths /home/deathmist/mesa/builddir/src/gallium/targets/dri, suffix _dri)", the radeonsi_dri.so definitely exists
16:54robclark: anyone have opinions on PIPE_QUERY_SO_OVERFLOW_ANY_PREDICATE query-buffer result, ie. does it *have* to be zero or one, or is zero vs non-zero good enough? Piglit seems to take the latter view, glcts the former. The latter is easier to implement..
16:56robclark: oh, I guess I already have to do the same gymnastics for occlusion predicate..
17:01zmike: what does the spec say
17:03robclark: "if the number of bits is non-zero, the minimum number of bits is 1" ... which kinda implies a zero/non-zero result is ok.. but I'm no spec lawyer
17:03robclark: from https://registry.khronos.org/OpenGL/extensions/ARB/ARB_transform_feedback_overflow_query.txt
17:04robclark: I can just throw an extra CP_COND_WRITE at it.. but a bit silly
17:17marex: is there any way I can still build mesa 23.1.y with meson 0.53 ?
17:18marex: (no, I cannot upgrade to newer meson, because that literally breaks build of almost every other component in the system)
17:21DavidHeidelberg[m]: marex: venv and pip?
17:21DavidHeidelberg[m]: marex: also meson upgrade itself hardly break anything anywhere I saw
17:25marex: DavidHeidelberg[m]: OE dunfell
17:27marex: DavidHeidelberg[m]: the meson 0.53 is fixed part of the build, updating that is problematic at best
21:35Lynne: could someone review !23227? we're just about to push the vulkan video code in ffmpeg, and rather than having something that often breaks, we'd rather have something that almost works