00:27karolherbst: imirkin: old monitors are just that bad :p
00:28imirkin: some of it is the backlight getting old
00:28karolherbst: but I kind of were using 150dpi displays for nearly 8 years now and using my not that bad 20" full HD display is always bad
00:29imirkin: well, same resolution as the old monitor
00:29imirkin: (for 24")
00:29karolherbst: uhh 1920x1200 at 24" is quite bad
00:30imirkin: i've never felt that way
00:30karolherbst: at those sizes 4k and even 2560x is more relevant
00:30karolherbst: makes quite the difference
00:30imirkin: maybe with fancy-graphics
00:32karolherbst: even with normal desktops it matters
00:32karolherbst: at some point you don't see pixels anymore
00:32karolherbst: and to go back above this threshold kind of hurts
00:33karolherbst: any kind of graphic anti aliasing sucks compared to having higher dpi natively anyway (even for fonts)
00:35karolherbst: had to use a 1680x1050 resolution on a 15" display once, and this was already terrible compared to 1920x1080
00:35imirkin: you get "into" the matrix? :)
00:35imirkin: i remember the first 14 or 15" LCD's at 1024x768
00:35imirkin: and those were 4:3 too, so you got more area per "inch"
00:36karolherbst: but there you could clearly see every single pixel
00:36imirkin: esp the stuck ones
00:36imirkin: i remember stores would only let you return it if it had more than 5 or something
00:36karolherbst: well, I was able to unstuck quite a lot of them though
00:37karolherbst: well, still the same today
00:37imirkin: yeah, but now it happens a lot less
00:37imirkin: i've bought a few dozen of these U2415's at work, and i don't think a single one had a stuck pixel
02:29imirkin: well that didn't take long... what's the way out of this? https://hastebin.com/ufigorituq.css
02:31imirkin: allllllrighty then. https://hastebin.com/hohoqociwu.cs
02:31imirkin: time to reboot into 4.18.12 and hope that lyude's changes have improved matters
02:31HdkR: Ace Ventura is here?
02:31imirkin: is that really what it's from?
02:32imirkin: i saw that move when i was like 10
02:32imirkin: [i guess that dates me, huh]
02:32HdkR: The slinky scene destroyed me as a child
02:32imirkin: o well. bbiab.
02:40imirkin: Lyude: messing with the mst status of monitors doesn't cause my xf86-video-nouveau branch to crash
02:41imirkin: as the connectors appear and disapper, nouveau thinks they're connected but they're really not
02:41imirkin: or rather, thinks they're on but they're not
02:43imirkin: Lyude: do you remember what situation you were having trouble with?
02:43imirkin: maybe something changes in Xorg 1.20? dunno
02:43imirkin: or maybe some userspace tool is the one responsible for various nastiness?
02:55imirkin: ok, well modesetting does the same thing -- when the monitors reappear, they're in dpms off, but xrandr thinks they're on
02:58imirkin: ah there ya go. a crash. no hotplug necessary.
04:18imirkin: heh. first time i see this! https://hastebin.com/bazemalowu.shell
04:36imirkin: skeggsb: should audio over DP-MST work?
04:36skeggsb: unlikely, i think it works a bit differently to sst
04:36skeggsb: never looked at it though
04:36imirkin: ok. i'm not seeing the ELD set
04:37skeggsb: yeah, that definitely won't be.. you're welcome to try the same stuff we do for DP audio, but iirc it needed different code
04:38imirkin: i'm not too fussed
04:38imirkin: the DP monitors' support for audio is minimal
04:38imirkin: they just have an audio jack that has no volume control
04:39imirkin: skeggsb: have you ever seen vblanks stop working all of a sudden?
04:39imirkin: in my case, it stopped randomly, i restarted glxgears, and it was fine again
04:39imirkin: i guess i'm not 100% sure it was vblanks...
04:39skeggsb: yeah, no
04:41imirkin: well, i made a few minor crash fixes to my xf86-video-nouveau code
04:41imirkin: but feature-wise, i don't see what people were complaining about =/
04:41imirkin: i tried the various cases, it all seems to work
04:41imirkin: connectors appearing, disappearing, etc
04:41imirkin: reappearing (sorta works - i think it needs some external intervention to do an actual modeset)
04:42imirkin: modesetting is like that too, so i figure it's ok
04:45imirkin: i think i'm gonna ship it
05:19imirkin: karolherbst: lmk if you plan on reviewing my mesa changes on list. if not, i'll push in the morning
21:25pendingchaos: imirkin: gl-3.1-vao-broken-attrib seems to work fine with c24c3b94ed2 ('gallium: decrease the size of pipe_vertex_buffer - 24 -> 16 bytes', when is_user_buffer was created)
21:25pendingchaos: git bisect seems to show 19a91841c34 ('st/mesa: Use Array._DrawVAO in st_atom_array.c.') as the breaking commit
21:25pendingchaos: haven't looked into it any further than that yet