00:06 alyssa: I could've sworn I wrote this patch a year ago.. https://rosenzweig.io/0001-mesa-st-Don-t-set-seamless-cube-maps-in-GLES2.patch
00:25 alyssa: meh.. advertising gles3 for deqp is easier :p
00:26 alyssa: looks like it's needed for true gles2 hw though
00:26 alyssa: anarsoul: ^^
00:26 anarsoul: ?
00:27 alyssa: anarsoul: dEQP-GLES2.functional.texture.mipmap.cube.basic.linear_linear fails on lima
00:27 alyssa: according to .gitlab-ci
00:27 alyssa: that patch probably fixes it
00:28 anarsoul: I see, thanks!
00:28 anarsoul: enunes: rellla ^^
07:41 rellla: alyssa, anarsoul: thanks, i'll look at it
21:06 solidhal: Hey dri-devel folks. Anyone here able to help me out with the dri-devel mailing list? I have a patch for an issue on the analogix bridge but my mail back as undeliverable, blocked by zen.spanhaus.org. I checked the block lists, but my domain is not on them. I also tried to email the list owner, but got thee same undeliverable reply. I have send
21:06 solidhal: mail to other kernel mailing lists before without much issue.
21:18 tlwoerner: solidhal: maybe create a PR on gitlab.freedesktop.org instead? i think that's the preferred workflow for patches
21:19 emersion: tlwoerner: not for the kernel
21:20 tlwoerner: emersion: haha, oops
21:20 emersion: :P
21:21 emersion: solidhal: what is your domain name?
21:21 solidhal: Thanks for the suggestion though tlwoerner :D
21:21 emersion: actually, can you send me an email about it? i'll try to have a look in the next few days
21:21 emersion: contact@emersion.fr
21:21 solidhal: emersion sure, thanks
21:24 solidhal: sent
22:17 airlied: zmike: sounds like yet another missing zink transition :-P
22:17 zmike: airlied: more like missing wsi
23:02 jljusten: tanty, anholt: Is there a simple thing I can run (locally) to test how we use renderdoc via piglit for Mesa's gitlab-ci?
23:03 anholt: jljusten: there's a readme in the piglit replay stuff
23:03 anholt: I also, for testing CI, will sometimes do a apt update; apt install -y <whatever> or wget https://people.freedesktop.org/~anholt/something in the job script to hack the job without rebuilding a container.
23:04 jljusten: anholt: so, one of the commands from piglit:replayer/README.md?
23:05 anholt: yeah. note that I don't have much experience with the new thing, I only reviewed a couple commits at the end of the series.
23:42 jljusten: hmm. I guess it's working. The sample `capture trace` command passes about 50% of the time if I use LIBGL_ALWAYS_SOFTWARE=1.
23:42 jljusten: I mean the 50% part is weird, but I think renderdoc is running to make the png in the results dir.
23:43 jljusten: All the pngs output there visually look similar. I'm not sure if there is a ref image to visually inspect.
23:51 anholt: I argued for checking in ref images, I was overruled.
23:52 anholt: you can go dig through pipelines on their dashboard to find other drivers for reference