07:05 K900: Hey folks, kind of a random question, but is there a specific reason Mesa doesn't read /etc/drirc.d?
07:05 K900: Just looking at https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/src/util/xmlconfig.c#L1266
07:05 K900: Just had a user lose some time to this being weirdly inconsistent
07:06 Ermine: is DATADIR /etc ?
07:06 Ermine: Ah, got the question
07:07 K900: No, DATADIR is /nix/store/aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa-mesa/share
07:07 K900: Which is read only and should not be touched
08:49 Company: is GFX12 still a nonconformant Vulkan implementation?
08:49 Company: because apparently GTK apps print a warning about that
08:50 K900: It's Mesa that's printing it
08:50 K900: Not GTK
08:50 pendingchaos: yes, that basically just means it doesn't pass ~100% of CTS
08:50 K900: So presumably yes
08:52 Company: it's GTK printing it because we connect to the debug callback
08:52 Company: and print warnings
08:52 Company: unlike Mesa, which turns them off by default
08:55 K900: Well Mesa is telling the debug callback to print it
08:55 K900: And presumably Mesa does know when it's conformant :)
08:56 Company: i'd just thought I'd point out that detail, because I've had arguments about printing those warnings to users by default
08:57 pendingchaos: not sure what debug callback you're talking about, vk_warn_non_conformant_implementation() is just does fprintf(stderr)
08:58 Company: oh, that isn't even the debug callback
08:59 Company: I take back what I said then - I had assumed it goes through _mesa_warning()
10:27 austriancoder: gfxstrand: where can I find the layering rules in vk runtime regarding interaction with wsi and actual drivers?
13:05 jani: anyone know how to get a full backtrace out of a kunit run? this is all I get with --raw_output=all: https://paste.debian.net/1412940/
13:06 jani: or could I be missing a kconfig option? which one?
13:31 glehmann: can someone explain to me why channel[0] is non void instead of channel[3] in util_format_description for PIPE_FORMAT_A8_UNORM?
13:32 glehmann: is that just a bug or am I misunderstanding and channel index 0123 is not rgba?
14:12 MrCooper: glehmann: that's a single-channel format, so only channel[0] exists
14:25 glehmann: yeah I think I actually want to look at the swizzle, not channels
16:28 tjaalton: dcbaker: your gpg key has expired 5y ago
16:31 tjaalton: ah, my cache was old, nevermind
17:25 dcbaker: tjaalton: no worries, my key does expire every couple of years and I usually forget to update it, lol
17:26 tjaalton: the debian tooling does fail though, as the primary key is "not live"
17:27 tjaalton: while here, how do I force meson to find a renamed bindgen?
17:32 elibrokeit: tjaalton: can be configured via https://mesonbuild.com/Machine-files.html
17:32 dcbaker: tjaalton: adding `[binaries] bindgen = '/path/to/bindgen'` to a native file should wok
17:33 elibrokeit: haha
17:33 dcbaker: apparently you get no answer, or too many answers
18:03 tjaalton: okay thanks, I'll give it a go