libera/#clim - IRC Chatlog

12:05:57 decweb Naive question: is McClim + SDL3 a thing?

12:07:25 jackdaniel I didn't touch sdl3 so far, so no

12:08:43 decweb tkx

12:09:16 decweb (had to ask based on your mastodon SDL2 dpi note)

12:11:21 decweb The inability to change DPI with the Wayland backend (after starting an app) is a source of frustration, so your SDL dynamic DPI thing was of interest, if I understood the post correctly.

12:12:27 jackdaniel is there a reason why you ask about sdl3, or simply you ask about thevnewest version?

12:12:59 jackdaniel we do our own glyph rendering, so we may assume any value for dpi

12:13:18 jackdaniel not necessarily thd real one taken from the screen

12:21:44 decweb My only reason is that the Fedora distro is sdl2_compat or some such thing, favoring sdl3.

12:23:39 decweb (with better words, sdl2 deprecated in favor of sdl3 on fedora, is my read on it, but I am not knowledgeable on this, thus the questions)

12:23:56 decweb meanwhile, asked and answered, and thanks.

12:24:47 decweb Btw I checked the backends page on the codeberg wiki, but it doesn't discuss sdl at all.

12:30:31 jackdaniel not merged to master yet

13:04:40 cmack decweb: are you talking about the wayland-ffi backend I'm working on or the sdl2 backend on a wayland compositor? If it's the former, I'm still actively working on fonts there...

13:06:46 decweb I was mostly interested in the sdl2 backend announcement because the mastodon-posted SDL2 DPI work caught my eye.

13:08:02 decweb But Fedora distros tend to clear the table while I'm eating, so there's always this little nag of using things that are no longer the active direction of Fedora.

13:08:26 decweb Once they try to ask for age verification I'll no doubt find a new distro :-)

13:08:52 cmack jackdaniel: from wayland-ffi perspective, I've made the graft retain the wl_output info, which includes the display scale.

13:11:01 cmack I haven't looked in a while, but I hope there is a wayland event when a surface (mirror sheet) moves to another output (graft)

13:14:25 jackdaniel cmack: display scaling and font dpi are are two different things though

13:15:13 jackdaniel most notably sometimes dpi is used to make font larger, while all dimensions are still expressed in "native" pixels

13:16:29 jackdaniel regarding my suggestion, that dpi could be stored with a graft, the main issue is that text-style-mapping takes a port as an argument, not the graft

13:16:52 jackdaniel but in practice, if we may have multiple ports of the same type (but with different initialization options), this is not /that much/ of on issue

13:18:40 cmack jackdaniel: yes, admittedly I haven't tackled anything beyond 1x, but previously I had thoughts on display dpi informing font dpi and whether to give override options.

13:20:09 jackdaniel I'm not saying that it is impossible, but would require at least a kludge. for example it could be that the font returned from the mapping returns a stand-in with defalut dpi, while medium-draw-text* consults graft and updates the cache with new glyphs depending on dpi

13:21:58 cmack jackdaniel: I maybe cheating with SDF fonts though since they are scalable... I'd only cache say the 32pt size and then scale its quad.

13:22:51 cmack however the TTF-using gl medium is now looking the best... so, I might already be on a collision course with that anyway

13:28:37 jackdaniel I'm kind of tempted to also add a sketch for hardware accelerated 2d medium (using sdl2 drawing primitives) and an incomplete opengl medium for experimentations; but I honestly shouldn't delay things further

13:30:28 cmack agreed. alpha warning is good enough now, IMO. I didn't mean to trigger more feature creep when requesting the destroy-port work :D

13:30:59 jackdaniel no no, it was important

13:31:01 cmack but that port arguments addition to master is very intriguing.

13:31:14 jackdaniel not being able to restart port woudl be a pita that requires restarting whole process

13:31:30 jackdaniel port arguments?

13:31:32 jackdaniel ah, find-port

13:31:37 jackdaniel well, we didn't behave as specified

13:31:54 jackdaniel and it is genuinely useful, because otherwise we wouldn't be able to pass initargs to ports added with find-port

13:32:05 cmack I've been wondering if I could so something like server-path = '(:wayland-ffi :renderer 'fixed-function)

13:33:57 jackdaniel you may destructure the path in initialize-instance

13:34:11 jackdaniel and access it, why not