freenode/#clim - IRC Chatlog
Search
13:34:37
loke
Another thing: I noticed that the existing code records the output record dimensions for text with a HEIGHT of (+ font-ascent font-descent). That menas that every output record with text of a given font size will _always_ have the same height. Regardless if the drawn string is just a single period, or a really high letter like "I"
13:35:01
loke
Is there any code that would rely on this fact? It doesn't seem the spec mandates this behaviour.
13:59:51
loke
This is some of the ugliest code I've ever written. Please tell me there is a nicer way to do this: https://gist.github.com/lokedhs/d53a0a691695561d3f5805c2e943963a
14:02:20
jackdaniel
loke: maybe (clim:transform-rectangle* clim:+identity-transformation+ left top right bottom) ?
14:06:10
loke
I wrote an alternative version: https://gist.github.com/lokedhs/d53a0a691695561d3f5805c2e943963a
14:11:24
jackdaniel
if there is such function I'd use it (if you are dissatisfied with the current code)
14:35:05
loke
Committed: https://github.com/lokedhs/McCLIM/commit/1e1abf7a6b18942cdc426f207148c3baf8eadd1e
14:53:58
loke
Now that I have a better font renderer, I've been able to make some progress on Climaxima: https://photos.app.goo.gl/1lv11fEGG01QIt5B3
15:00:16
loke
I'm tempted to go down yet another rabbit hole; to implement A lines drawing. And that would be A with subpixel sampling of course, because I'm a machosist
15:07:34
loke
subpixel? Well, kinda. It does work for colour, but I believe the algorithm is messy. I'm juts letting xrender and freetype handle it for me.
15:10:10
jackdaniel
imo optional antialiasing would be really good for mcclim. right now arrows which are not aligned to xy axis for that instance look atracious
15:11:21
loke
Ass we habve to do is to use xrender for drawing lines instead of the older Xlib primitives
15:14:47
loke
Drawing text using xrender seems to be harder than drawing lines, and I've already done text :-)
15:20:24
loke
One can also draw a monochrome AA arc in memory, and blit it using xrender. I'd imagine that to be rather inefficient htough.
15:23:49
loke
It needs better configurability, but that will be improved once I get my configuration framework done.
15:30:00
loke
There is certainly more hacking I can do on it, and I'm in no particular hurry to merge
15:37:31
fittestbits
jackdaniel - watched your video, looks good. I'm looking forward to where you go with it.
15:41:11
jackdaniel
loke: unless someone else reviews your code I'll go over it after long weekend if that's fine. are there any instructions how to activate/deactivate it? or some summary of what has been changed?
15:44:16
loke
The current code autoamtically activates it (there is a SETQ in the main file). It's activated by a :FONT-RENDERER option in the path.
19:59:23
jackdaniel
I consider putting the videos as a playlist (so pieces vary between 10 and 30 minutes) instead of posting one over-2h-video
21:40:55
makomo
jackdaniel: that would be ok if the videos are logically separated. if they aren't, maybe just a list of timepoints in the description could do