06:37dviola: ehh.. not really, doesn't look like the bug is fixed, was just not easily reproducible
16:23wlb: weston Issue #265 closed \o/ (Pipewire output is always two frames late https://gitlab.freedesktop.org/wayland/weston/-/issues/265)
16:57wlb: weston Issue #711 closed \o/ (Clipboard doesn't appear to working when using in-tandem with the screen-share module https://gitlab.freedesktop.org/wayland/weston/-/issues/711)
17:23wlb: weston Issue #756 closed \o/ (cleanup after cairo / xwayland / font map hash table destroy https://gitlab.freedesktop.org/wayland/weston/-/issues/756)
17:24Consolatis: dviola: I see the same issue for an old version of VLC running via xwayland. Basically all fonts are gone on output-loss. For some reason the playlist entries are the only ones remaining. Even the menu entries have no text. I am also not quite sure how to reproduce the issue, I think it has something to do with focus at the time of output loss but didn't feel like debugging so far
18:14dviola: Consolatis: I see, I found this bug report which seems to be the one that is similar to my issue: https://bugreports.qt.io/browse/QTBUG-108602
18:14dviola: it also links to a keepassxc bug report
18:23Consolatis: yep, that seems like my VLC issue as well. The freezing thing I only get sometimes and only when trying to close the VLC window after the fonts are gone though, I can still interact with VLC otherwise
18:45dviola: I see, looks like it's fixed in qt 6.x
18:46dviola: I thought about bisecting old qt 5.x to see where it broke
18:49dviola: I hear they could still backport stuff to https://invent.kde.org/qt/qt/qtbase/-/tree/kde/5.15?ref_type=heads but I'm not sure
18:54dviola: I was also advised not to get near 5.x which seems to go well with my experience when trying to build it last time
18:55dviola: s/get/go/
19:21colinmarc: I noticed that vkcube-wayland started using llvmpipe with my compositor on nvidia proprietary 560 drivers. Does the new nvidia WSI require explicit sync or any other protocol to be present?
19:36colinmarc: eglinfo has no problem, so I think it must be specific to their vulkan WSI
19:49bl4ckb0ne: I just hit this with SDL3 https://gitlab.freedesktop.org/mesa/mesa/-/issues/10254 wouldn't it be possible to at least VK_SUBOPTIMAL_KHR if the buffer and the viewport size doesn't match?
21:12DemiMarie: Would it be possible to get a review on https://gitlab.freedesktop.org/wayland/wayland/-/merge_requests/444?
21:13DemiMarie: Right now any client can cause the server to spin and consume CPU.
21:23ManMower: DemiMarie: might be worth adding a few CCs on that ticket. the people that wrote/reviewed the original breakage might not be paying attention closely.
21:24DemiMarie: ManMower: do I make an new comment or edit the original?
21:24DemiMarie: ManMower: thank you!
21:24ManMower: oh make a new comment
21:24ManMower: if you just edit it won't fire off emails :)
21:24DemiMarie: Will do.
21:24ManMower: thanks!
21:25DemiMarie: I have a few other fixes coming as well.
21:26ManMower: cool
22:19mclasen: DemiMarie: I think we very much want to allow messages that are larger than 4k
22:20DemiMarie: mclasen: that is going to be a much larger fix and it never worked before.
22:22DemiMarie: Rejecting such messages is the right thing to do for now.
22:32DemiMarie: mclasen: If clients send messages longer than 4k a server using libwayland 1.22 will disconnect them, and a server using libwayland 1.23 will busy-spin. If longer messages are needed, they should be explicitly negotiated.
22:44mclasen: spinning isn't desired, obviously