ChanServ changed the topic of #zink to: official development channel for the mesa3d zink driver || https://docs.mesa3d.org/drivers/zink.html
pendingchaos has quit [Server closed connection]
pendingchaos has joined #zink
hikiko has quit [Server closed connection]
hikiko has joined #zink
LexSfX has quit [Remote host closed the connection]
LexSfX has joined #zink
LexSfX has quit [Remote host closed the connection]
LexSfX has joined #zink
<kusma> zmike: It seems kinda fails to render Portal 2 correctly, displaying a bunch of meshes white... Something you're aware of?
<kusma> Ticket incoming from the person who actually discovered it, I just reproduced it from a private discussion ;)
<zmike> kusma: I don't remember if I was aware
<kusma> I mean, seems Portal 2 has a vk renderer, so I guess that game isn't of particular interest to you.
<zmike> it has dxvk
<zmike> not actual vulkan
<kusma> Ah
<zmike> we can have a ticket anyway though to track it
<zmike> might take a look if I have time
<kusma> Yeah, I was told by Italo that he'd file one soon
<zmike> cool
<zmike> have you had a chance to look at the cube thing again?
<kusma> He's working on Zink + Venus BTW
<zmike> oh no
<kusma> No, I'm right now a bit busy with some client work, sadly.
<zmike> that is sad
<kusma> I'll try to take a look tomorrow, though.
<zmike> thanks
<zmike> branchpoint is in 2 weeks and I'd like to get that fixed and landed
<kusma> Yeah, I get that.
LexSfX has quit []
LexSfX has joined #zink
<ajax> zmike: copper updated with rbs on everything. still have a crash to sort before ci believes in it.
<zmike> ajax: noice
<zmike> xcb_wait_for_special_event is really proving the bane of our existence these days
<ajax> it's such a bad api
<ajax> i've thought about adding window destroy notification events to Present just so i can be sure we always get a go-away special event
<ajax> and when you're revving the wire protocol to work around your library's API...
<zmike> would be nice
<zmike> haha
<ajax> so the bug here is if you use EGL_EXT_device_blah to explicitly pick the software device. on main you'd get llvmpipe, on copper it kindaloads zink but crashes.
<zmike> hmm
<zmike> seems not ideal
<ajax> think we should just keep that on lp unless you've done the usual MESA_LOADER_DRIVER_OVERRIDE
<zmike> probably
<zmike> the intent with current stuff is that the env var must be specified to get zink
<ajax> we don't expect kopper to work for EGL_MESA_platform_gbm yet, right?
<ajax> just x11 and wayland
<zmike> uhhh
<zmike> I dunno?
<zmike> I think it worked for that when I was testing weston?
<ajax> weston as the top-level compositor? eesh.
<zmike> yeah I blogged about that working last year
<zmike> so at some point it probably worked
<ajax> hmph
<zmike> nice try :D
<ajax> okay well hm
<ajax> i'm not sure how to easily keep that working post-kopper
<zmike> I had it working with kopper though
<ajax> oh?
<zmike> that's what I blogged about
<zmike> so at some point it worked
<ajax> the blog post doesn't mention kopper but i suppose the timing makes sense
<zmike> I remember hacking the gbm stuff up for that too
LexSfX has quit []
LexSfX has joined #zink