01:11wlb: wayland-protocols Merge request !279 opened by Poly (PolyMeilex) xdg-shell: Introduce set_modal and unset_modal https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/279
11:21pq: dubiousness, a HMD is a view into a virtual 3D space. Positioning in that can be anything, really. Posters on many walls, not necessarily straight. Also non-linear transformations, like projection on a dome. In that case, the surface local coordinate axes could be curved in 3D.
12:59dubiousness: pq: sure, so I would imagine it would be up to the compositor in that case to say: This is a wall, here are the coordinates; this is a dome, here are the details; this is a discarded pizza box, etc.
13:00dubiousness: so that anything “launched” and I use that word carefully, would be relative to the surface it’s presented with, to avoid egregious things like arbitrary overlapping
13:13wlb: wayland-protocols Merge request !279 closed (xdg-shell: Introduce set_modal and unset_modal)
13:14dubiousness: hmm, no, I think I’m incorrect
13:57pq: dubiousness, those were just a couple obvious examples, I'm sure there are infinitely more choices.
14:15Company: fractional scaling question
14:15Company: on a 150% surface with an odd width - say 11px - I am supposed to create a round (11 * 1.5) = 17px buffer, right?
14:16Company: and am I supposed to render into it as if the buffer represented 11px or as if the buffer represented 17/1.5 = 11.333333px?
14:17Company: or am I supposed to specify that myself using the viewporter source rect?
14:18Company: ie setting destionation = 11; source = 16.5px?
14:54bl4ckb0ne: dubiousness: i have https://git.sr.ht/~bl4ckb0ne/wxrc/tree/master/item/protocol/xr-shell-v1.xml and https://git.sr.ht/~bl4ckb0ne/wxrc/tree/master/item/protocol/xr-input-v1.xml that ive wanted to upstream for ages if that can help you
14:54dubiousness: bl4ckb0ne: ack, I will have a look
14:55bl4ckb0ne: ping me if you have any questions
17:12d_ed[m]: <Company> "on a 150% surface with an odd..." <- yes
17:13Company: to all the questions?
17:14d_ed[m]: to the rounded buffer
17:15d_ed[m]: and then setting the source rect to the whole buffer, destination to 11
17:16Company: I had so far assumed that keeping source unset would do the right thing
17:16Company: but I suppose it doesn't for the odd case
17:17Company: does the math for this work out when maximized fullscreen?
17:17Company: 1920x1080, 175% scale factor
17:18Company: 1920/1.75=1097.14...
17:18Company: Do I get a size of 1097 in that case?
17:19d_ed[m]: sure, unset is the same thing
17:19Company: but whatever I do, I'm gonna be screwed
17:20Company: this is gonna be so much fun
17:20d_ed[m]: It certainly has it's share of issues
17:21wlb: wayland Merge request !366 merged \o/ (event-loop: use wl_priv_signal for the destroy signal https://gitlab.freedesktop.org/wayland/wayland/-/merge_requests/366)
17:21wlb: wayland/main: Kirill Primak * event-loop: use wl_priv_signal for the destroy signal https://gitlab.freedesktop.org/wayland/wayland/commit/6a7284c63234 src/event-loop.c
17:21d_ed[m]: Company: that's what we do in Qt and SDL at least
17:23Company: 3836x2398 is the buffer I end up with on GTK when fullscreening on 175%
17:23Company: there's more than 1 thing going wrong when I'm off by 2
17:25Company: 125%, 150% and 250% come out right
17:25Company: at least
17:25Company: that's why you don't use something with a factor 7 I guess
17:26Company: 3834x2397
17:26Company: at 225%
17:26Company: oh well
17:41d_ed[m]: I tested Qt + kwin on a 4k monitor @ 175% with debug on the client... (full message at <https://matrix.org/_matrix/media/v3/download/matrix.org/jcdyocIaUnBpisLfLcaSDwHC>)
17:48Company: yeah, something is going wrong here
17:48Company: 175% should work
17:48Company: though you can theoretically get in situations where you can't round to the correct number anymore once the scale is > 2
18:09zamundaaa[m]: You don't need scale > 2 for that
18:10zamundaaa[m]: It happens plenty with lots of scale + size combinations in between scale 1 and 2
18:21Company: I would have expected you can always send a size that rounds properly, but I guess if the client always rounds up, you're indeed right
18:21Company: just because there's fewer numbers between [0, x/scale] and [0,x]
18:22Company: so there can't be an injective function to map between them
18:23Company: though I'm mostly interested in the default scales - 125% 150% 175% - and the default monitor sizes
20:27kennylevinsen: Company: what's the configure dimensions you get from the compositor? For your 3840x2400 display it should be 2194x1371... but you'll only get to 3840x2339 due to rounding.
20:31Company: 2192, 1370 apparently
20:32Company: bad mutter
20:53orowith2os[m]: whot: any particular reason for the stalling of https://gitlab.freedesktop.org/libinput/libinput/-/merge_requests/426 for so long?
20:54orowith2os[m]: the general idea seems good; make it simpler
20:55orowith2os[m]: actually, I should take a look at the modern source, make sure it still matches up, so I'm not running off an old diff....
20:56orowith2os[m]: okay, yeah, still about the same