02:02wlb: wayland Issue #375 opened by Cahal Dallett (Doorknob4746) Wayland resolution options for Sony A8H TV https://gitlab.freedesktop.org/wayland/wayland/-/issues/375
05:22slattann: test Msg
07:52pq: bluebugs, libwayland-client simply has no event loop machinery, so it has not add_idle either.
07:53pq: slattann, test ok.
07:53pq: bluebugs, sorry, I meant bl4ckb0ne
07:59wlb: wayland Issue #375 closed \o/ (Wayland resolution options for Sony A8H TV https://gitlab.freedesktop.org/wayland/wayland/-/issues/375)
09:42slattann: @pq: Explicit Synchronization MR: https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/90
09:42slattann: What are the next tasks pending on this get merged ?
09:51emersion: get the kernel folks to decide whether they want to throw away drm_syncobj
09:52MrCooper: slattann: out of curiosity, why do you want this?
10:00pq: slattann, sorry, I try hard to concentrate on color management, so I don't want to spread to that topic for now. Or VRR or...
10:39wlb: weston/main: Leandro Ribeiro * tests/color-icc-output: differentiate test name based on ICC profile type https://gitlab.freedesktop.org/wayland/weston/commit/7ea4075d777c tests/color-icc-output-test.c
10:39wlb: weston/main: Leandro Ribeiro * tests/color-icc-output: assert that dimension is not zero when creating clut https://gitlab.freedesktop.org/wayland/weston/commit/722d7f8c3f27 tests/color-icc-output-test.c
10:39wlb: weston/main: Leandro Ribeiro * tests/color-icc-output: add ICC VCGT tests https://gitlab.freedesktop.org/wayland/weston/commit/8c9dd4febbe1 tests/ reference/output-icc-decorations-03.png reference/output-icc-decorations-04.png reference/output_icc_alpha_blend-03.png reference/output_icc_alpha_blend-04.png reference/shaper_matrix-03.png reference/shaper_matrix-04.png color-icc-output-test.c col
10:40wlb: weston Merge request !1226 merged \o/ (Add ICC VCGT tests https://gitlab.freedesktop.org/wayland/weston/-/merge_requests/1226)
11:10pq: emersion, did you have anything in mind I should comment about of the series https://lore.kernel.org/dri-devel/20230411222931.15127-2-ville.syrjala@linux.intel.com/ ? It seems ok to me, but I wouldn't know if e.g. the row-major layout is correct.
11:11emersion: pq, ah, i was hoping you would know that
11:13pq: I've never used a KMS CTM before.
11:13emersion: i see
11:13emersion: JoshuaAshton: have you by any chance? ^
11:13pq: I do like the layout documented very very much if it's correct. :-)
11:14pq: and would probably be one of the first reviee comments if I was reviewing a patch adding that API now
11:15pq: btw. you might like https://github.com/mm2/Little-CMS/issues/377
11:26emersion: nice!
12:17zamundaaa[m]: pq: I'm using the CTM and the order is correct
12:21pq: zamundaaa[m], would you like to send a R-b to that patch?
12:33zamundaaa[m]: sure
12:35pq: thanks!
13:16bl4ckb0ne: pq yeah i understand the reasons behind that choice
13:17bl4ckb0ne: but having a way to add idle tasks from the client side could save me some headache
13:17pq: bl4ckb0ne, that requires an implementation of an event loop, so.
13:18bl4ckb0ne:puts a todo for an event loop in his project
13:18pq: FWIW, weston's clients/window.c has a simple one IIRC. But maybe it would be better to use an actual library.
13:18pq: next you want a timer
13:18pq: then to watch some other fd
13:21bl4ckb0ne: ill take a look at that file, thanks
13:23pq: display_run() has the loop, display_defer() is the idle task
13:26pq: there is even toytimer_*(), but that is wasting timer fds
13:37bl4ckb0ne: the deferred task list might be what i need
13:38kennylevinsen: that reminds me of the "unentangle libwayland-server from wl_event_loop" task...
18:50any1: libevent, libev, libuv, aml. Why not just pick one? :)
18:55kennylevinsen: I'd rather have the compositor (or at least compositor lib) have the freedom to pick...
18:56kennylevinsen: it's not like libwayland needs deep event loop integration just to poll a few client sockets
18:57emersion: yea
18:57emersion: i have a MR but pq doesn't like it too much
18:59kennylevinsen: The issue was the complete removal of wl_event_loop, right?
19:01kennylevinsen: Maybe we could convince him by starting softer, keeping it but not using it.
19:02kennylevinsen: Better than nothing, and I imagine it wouldn't be used by compositors for *too* long after that as there are much nicer options out there. Easier to go for the kill then than while everyone still use it...
19:11bl4ckb0ne: i really like that display defer idea
21:27JoshuaAshton: emersion, pq: This is what it looks like:
21:28JoshuaAshton: https://github.com/ValveSoftware/gamescope/blob/297245cd7ecc8cde53aba06d439825024d002541/src/drm.cpp#L2408
21:28JoshuaAshton: (CTM is currently not hooked up in main branch of Gamescope yet, pending me merging some kernel stuff for per-plane CTM for my scRGB stuff)
21:28JoshuaAshton: but that from the last proper release should explain it
21:31JoshuaAshton: pq: Re. scRGB I was saying to clip to 2020 before doing color mgmt like a typical HDR10 plane.
21:31JoshuaAshton: and as for HLG monitors, ngl I really don't care about supporting those, and given my discussions with movie + vfx people, neither does the rest of the industry anymore.
21:32JoshuaAshton: and yeah I would use the content primaries in metadata to determine what I do for color mgmt for scRGB content, same as HDR10.