00:38wlb: wayland-protocols Merge request !390 opened by Xaver Hugl (Zamundaaa) Draft: color: add an absolute colorimetric + black point compensation rendering intent https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/390
08:10wlb: wayland-protocols Merge request !391 opened by Jonas Ådahl (jadahl) build: Bump version to 1.43 https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/391
08:15wlb: weston Issue #1016 opened by A B (absn) No input devices on entering Weston https://gitlab.freedesktop.org/wayland/weston/-/issues/1016
08:27wlb: weston Merge request !1720 opened by Philipp Zabel (pH5) backend-drm: Work around amdgpu issue #1890 https://gitlab.freedesktop.org/wayland/weston/-/merge_requests/1720
10:53swick[m]: riteo: yes, only WSI supports fifo-v1 currently. WSI requires commit-timing-v1 as well for FIFO, so I think checking for those two protocols should be enough.
11:00zamundaaa[m]: It does not require commit timing
11:18wlb: wayland-protocols/main: Jonas Ådahl * build: Bump version to 1.43 https://gitlab.freedesktop.org/wayland/wayland-protocols/commit/4313a51a170e meson.build
11:18wlb: wayland-protocols Merge request !391 merged \o/ (build: Bump version to 1.43 https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/391)
11:47wlb: wayland-protocols New tag: 1.43 https://gitlab.freedesktop.org/wayland/wayland-protocols/tags/1.43
11:47davidre: \o/
11:52swick[m]: zamundaaa: it does for forward progress on some compositors, so its availability is required to be sure there is forward progress
11:59zamundaaa[m]: No, it does not require it
11:59zamundaaa[m]: If a compositor implements fifo only, it gives forward progress guarantees with that
12:07swick[m]: that was changed recently apparently
12:07swick[m]: so now instead of getting no forward progress on some compositors you get unthrottled progress
12:08swick[m]: which means that you still have to check both protocols to be somewhat confident that the behavior will be throttled with forward progress
12:08swick[m]: riteo: ^
12:08zamundaaa[m]: No, you do not get unthrottled progress
12:08zamundaaa[m]: If a compositor only implements fifo, you can assume it does it properly
12:09zamundaaa[m]: And throttles on the compositor side
12:09swick[m]: fifo explicitly says that it's okay to run unthrottled when the surface is occluded or otherwise doesn't have an associated refresh rate
12:09swick[m]: and it's what mutter does
12:09zamundaaa[m]: Yes, but mutter also supports commit timing
12:10swick[m]: yes
12:10wlb: wayland.freedesktop.org Merge request !97 opened by Jonas Ådahl (jadahl) releases: Add wayland-protocols 1.43 https://gitlab.freedesktop.org/wayland/wayland.freedesktop.org/-/merge_requests/97
12:10wlb: wayland.freedesktop.org Merge request !97 merged \o/ (releases: Add wayland-protocols 1.43 https://gitlab.freedesktop.org/wayland/wayland.freedesktop.org/-/merge_requests/97)
12:10wlb: wayland.freedesktop.org/main: Jonas Ådahl * releases: Add wayland-protocols 1.43 https://gitlab.freedesktop.org/wayland/wayland.freedesktop.org/commit/81c55299b7a6 releases.html
12:10swick[m]: sorry, but it's completely valid to just support fifo
12:10swick[m]: which has been one of my objections to the protocol
12:10swick[m]: but hey, you get what you get
12:10zamundaaa[m]: Yes, and you can expect it to work properly when that is the case
12:10swick[m]: yes, according to the fifo protocol
12:11swick[m]: which allows unthrottled
12:11zamundaaa[m]: If the compositor does unthrottled without also supporting commit timing, then that's the compositor's fault
12:11swick[m]: no, because the protocols allow for that
12:12zamundaaa[m]: Protocols also allow you to turn all client windows black. Doesn't mean it's something any compositor does
12:12swick[m]: no, but you also don't have the guarantee that all your client windows are not black
12:12swick[m]: you only get the guarantees that the protocols make
12:13zamundaaa[m]: Yes, so what's the point of checking for a completely hypothetical situation in which a compositor does something incredibly stupid?
12:13swick[m]: it's not hypothetical
12:13zamundaaa[m]: It very much is
12:14swick[m]: if the protocol says you can run unthrottled and you don't want unthrottled, you might get unthrottled
12:14zamundaaa[m]: The same way a compositor could send frame callbacks infinitely fast, fifo not being throttled when commit timing isn't available is not a practical concern
12:15swick[m]: like, if you're fine not having a guarantee then there is nothing you have to do, but if you want to have a guarantee, then you need to actually care about what the protocols allow and what they do not
12:15pq: mind the temperature guys, take a breath once in a while ;-)
12:16swick[m]: apparently you're fine not having guarantees, and that's totally fine, but if someone asks what they have to check to be sure that they get unthrottled forward progress with FIFO then the answer is that both protocols need to be supported, otherwise you can't be sure
12:19zamundaaa[m]: The fifo protocol doesn't guarantee anything. A compositor can still very much throttle when commit-timing is supported as well
12:21zamundaaa[m]: Even if the protocol did guarantee things, the protocol being supported doesn't necessarily mean the driver uses it too. In the end, apps just need to go with "most likely works", which is not good at all, but we just don't have anything better
14:25wlb: wayland-protocols Merge request !390 closed (Draft: color: add an absolute colorimetric + black point compensation rendering intent)
21:20riteo: swick[m], zamundaaa[m], thank you both for the clarification!
21:21riteo: If and when EGL will get forward progress, will there be a way for detecting that?
21:59wlb: weston/main: Robert Mader * gitlab-ci: Update Meson https://gitlab.freedesktop.org/wayland/weston/commit/9675ef49f20e .gitlab-ci/build-deps.sh
21:59wlb: weston/main: Robert Mader * gitlab-ci: Update Mesa for llvmpipe dmabuf support https://gitlab.freedesktop.org/wayland/weston/commit/57f48ab91cee .gitlab-ci/build-deps.sh
21:59wlb: weston/main: Robert Mader * gitlab-ci: Build kernel with udmabuf support https://gitlab.freedesktop.org/wayland/weston/commit/682d420b5010 .gitlab-ci/ build-deps.sh virtme-scripts/run-weston-tests.sh
21:59wlb: weston/main: Robert Mader * gitlab-ci: Bump image https://gitlab.freedesktop.org/wayland/weston/commit/81800294a98a .gitlab-ci.yml
21:59wlb: weston Merge request !1717 merged \o/ (gitlab-ci: Add udmabuf support and switch to swrast https://gitlab.freedesktop.org/wayland/weston/-/merge_requests/1717)