10:21 <emersion> oh, that was fast: https://oftc.irclog.whitequark.org/wayland/2021-05-24

10:22 <emersion> (also added #dri-devel and #freedesktop while at it)

10:23 <daniels> woot!

10:23 <pq> sweet

10:24 <emersion> hmm, irkerd isn't so great

10:25 <emersion> it requires irkerd to be hidden behind a firewall in the same network as gitlab's

10:26 <emersion> could always whitelist FDO IPs in my firewall, but it's a tad annoying if they're subject to change…

11:44 <daniels> ah fun, no authentication

11:45 <emersion> yeah :/

12:21 <pq> AbhijeetK, you also need to register and identify with nickserv to speak, because here too we suffer from spam. I mean, I haven't seen any message from you yet.

12:34 <AbhijeetK> Hi @pq

12:38 <pq> AbhijeetK, heya

12:39 <AbhijeetK> About deducing a fractional scale from `wl_output.mode/xdg_output.logical_size`, you suggested that a wayland compositor is allowed to do arbitrary, even non-linear transformations to the surface. Looking at isolated examples like that does not work.

12:39 <AbhijeetK> I am still not able to get why compositor will do that ? Because protocol always wants compositor to send standard transformation mentioned here -> https://gitlab.freedesktop.org/wayland/wayland/-/blob/main/protocol/wayland.xml#L2479

12:51 <JPEW> AbhijeetK: Because your compositor could be something like this: https://www.youtube.com/watch?v=_FjuPn7MXMs

12:54 <pq> or maybe the monitor is just a plane in VR space

12:54 <pq> observed via HMD

12:55 <pq> or picture-in-picture on something else

12:57 <pq> AbhijeetK, the wl_output transform is something else: it meant to allow direct scanout of client content.

13:01 <AbhijeetK> Thanks @pq

13:01 <pq> AbhijeetK, nothing in Wayland actually says that you cannot have a circular monitor with a pixel grid at a 40-degree angle, but Wayland does offer opportunities to take advantage of common cases where a more optimal hardware path is possible.

13:02 <pq> wl_output interface is also one of the oldest things, and doesn't really live up to the modern Wayland ideas

13:05 <pq> AbhijeetK, in general, one should not try to infer new things from old parameters that were never intended for it - like trying to guess the optimal buffer size from wl_output and xdg_output information.

13:07 <pq> If you need an optimal buffer size in a client, write a new protocol extension that delivers exactly that, with an explanation of what it actually means and what it is intended to be used for. Even doing that it will be hard to make everyone understand it the same way.

13:12 <AbhijeetK> pq, Make sense. Thanks for the explanation. I was always thinking about a rectangular monitor display. I should read more about it. Can you suggest some resources?

13:17 <pq> I don't know of any references.

13:18 <pq> desktop-in-VR has been tried in a few places I think, and smart watches are a thing

13:19 <pq> smart phones rarely have an exactly rectangular display either, there are rounded corners and carve-outs and curved-to-the-side displays

13:20 <pq> on portable devices, one may also animate an orientation change - that is a temporary arbitrary image transformations which you probably wouldn't want to clients to keep up with during it.

13:58 <pq> whee, I got test_viewporter_double_create to be ASan-clean. \o/ ... aside from the leaks inside Cairo.

13:59 <pq> that was a bit of a handful, since it uses desktop-shell, which spawns weston-desktop-shell and weston-keyboard and they all needed fixing, and toytoolkit.

14:09 <pq> fixed leaks, trigger new UAFs elsewhere

14:13 <daniels> https://xrdesktop.freedesktop.org/ does exist

14:15 <pq> it's very... terse

14:15 <daniels> that's ... impressive

14:16 <daniels> I saw the logo and just assumed you could scroll down and get, y'know, content

14:16 <pq> me too, or like something to click

14:16 <daniels> https://www.collabora.com/news-and-blog/news-and-events/moving-the-linux-desktop-to-another-reality.html works, since I know you're always asking if you can have demonstration videos and release [pv]odcasts instead of descriptive text

14:16 <jadahl> i pressed ctrl+h just to see if I was missing something

14:17 <JPEW> Maybe you have to view it with a virtual reality headset to get the true expirence

14:17 <pq> I definitely missed something in shell_output destruction

14:17 <pq> now it gets freed twice

14:17 <daniels> JPEW: 'it's behind you!'

14:18 <jadahl> hmm, I suggest that one isn't a person

14:18 <jadahl> is gitlab-bot the new wlb?

14:19 <daniels> I believe it goes in exactly the opposite direction

14:19 <daniels> e.g. if I type https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/8

14:19 <gitlab-bot> wayland issue (Merge request) 8 in wayland-protocols "linux-dmabuf: add wp_linux_dmabuf_hints" [Linux-Dmabuf, Opened]

14:19 <daniels> ... then that happens

14:19 <jadahl> ah, that one

14:20 <jadahl> because its natural to have one gitlab bot per direction

16:22 <swick> the whole fractional scaling thing seems to be running in circles

16:23 <glennk> isn't that fractal scaling then?

16:24 <swick> and it probably will continue doing so without someone sitting down and just implementing it

16:26 <swick> glennk: took me way longer to spot the difference than I want to admit

17:20 <lanodan> Took me many reads as well…

19:36 <silver> oi