06:25 phasta: Did we hear from Linus yet about the Link: tag we add to all patches? Someone told me he doesn't want that anymore
06:47 mlankhorst: From what I understand he only wants those links if it adds context, but I'm not sure how it affects drm. Personally I just keep adding those as long as dim adds those automatically. :-)
08:02 mripard: phasta: he changed his mind
08:03 phasta: mripard: He did?
08:04 phasta: I also think the links would be more useful if the series linked to contained discussions always. But making them optional in dim would lead to maintainers forgetting them when it does make sense, so…
08:22 mripard: phasta: yeah, some time last week
08:23 mripard: but he agrees to it if we use a separate domain for those that will be setup by Konstantin
08:23 mripard: so we can easily tell the difference between context and reference to the patches being posted
08:23 mripard: which is reasonable to me
08:29 mlankhorst: Would it be interesting to have a script in maintainer-tools to allow using thunderbird instead of mutt for DIM_MUA?
13:43 demarchi: mlankhorst: I don't use thunderbird myself, but it should be just a matter to adjust DIM_MUA in your dimrc, no?
13:57 mareko: Venemo: yes, but "extensions->ARB_compute_variable_group_size =" needs to be updated to >= 512
13:57 Venemo: mareko: ah, right, I see what you mean
14:08 mlankhorst: demarchi: I needed a parser for it. ;-)
14:09 demarchi: a parser? something to convert the command line options we use to what thunderbird uses?
14:10 mlankhorst: Yeah, a small getopt loop mostly
14:11 mlankhorst: Are all maintainers using mutt?
14:13 demarchi: mlankhorst: I'd rather change it to something using envsubst or a template string, so people don't have to create separate scripts for each email client
14:14 mlankhorst: demarchi: what mua are you using then?
14:14 demarchi: neomutt
14:57 jhugo: Did something happen to gitlab.freedesktop.org in the past couple of days that I missed? I seem unable to fetch repos via ssh
14:57 gfxstrand: are you fetching from ssh.gitlab.freedesktop.org?
14:58 jhugo: drm ssh://git@ssh.gitlab.freedesktop.org/drm/kernel.git (fetch)
14:58 gfxstrand: hrm... That's working for me
15:00 jhugo: Huh. Might be our firewall, although other ssh connections work
15:24 gfxstrand: dj-death: I'm not sure about the error handling in vkCreatePipelineBinariesKHR()
15:24 gfxstrand: The spec says: "
15:24 gfxstrand: The implementation will attempt to create all pipeline binaries. If creation fails for any pipeline binary, then:
15:24 gfxstrand: So should we really early return?
15:25 gfxstrand: Or should we just go "oops" and move on to the next one?
16:16 zmike: tarceri: I managed to pinpoint that weird hash table sizing regression to something in your wheelhouse
16:17 zmike: maybe you can take a closer look
16:31 dj-death: gfxstrand: what can you do with half a gfx pipeline? :)
16:33 alyssa: dj-death: capture stream out and then discard rast ;)
17:02 mlankhorst: demarchi: Seems from a quick poll, everyone is using a variation of mutt and neomutt, so it shouldn't on its own be a problem.
18:00 gfxstrand: dj-death: We can at least tell them what shaders were missing from the cache if they ask for internal caching
18:01 gfxstrand: dj-death: It just seems kind of mean to tell them that the pipeline has 8 binaries and then when they go to ask for them from the internal cache say "Oh, actually only 4 binaries and, BTW, something's missing"
18:01 gfxstrand: Like, if we claim 4 binaries, we should always return 4 binaries
18:01 gfxstrand: But maybe we were doing that?
18:02 gfxstrand: No, I'm very sure we weren't
18:24 alyssa: gfxstrand: counterpoint, it seems mean of the app to use pipelines ;P
18:27 gfxstrand: Or to expect binaries out of them, if we're being hones
18:27 gfxstrand: *honest
18:28 alyssa: Exactly
18:28 alyssa: ESO is fine
18:31 gfxstrand: Oh, well. Soon I'll have reviewed dj-death's MR and then we can all forget about pipelines again for a while.
18:33 alyssa: hopefully forever
18:33 gfxstrand: Oh, you'd like that, wouldn't you
18:33 gfxstrand: I'm sure we'd all like that
18:33 gfxstrand: And I'm sure I'll be typing pipeline code again in another 6 months
19:06 dj-death: gfxstrand: it's called pipeline_binaries not shader_binaries ;)
19:12 alyssa: dj-death: yes because ESO got this right first try
19:13 gfxstrand: You mean serializing pipeline caches wasn't "right"?
19:15 alyssa: *melt*