freenode/#clim - IRC Chatlog
Search
4:25:19
shrdlu68
In pages 6-7 of the clim user manual (pdf), I can't get that simple example to work. It seems (define-application-frame my-first-clim-app...) should define a class with the name my-first-clim-app, but the interpreter warns that it's an undefined variable.
4:31:26
beach
It sounds like it doesn't recognize that define-application-frame is a symbol in the CLIM package.
4:52:09
jackdaniel
loke: it won't be reproduced (most likely), because you have forked after the fix was applied. but the problem is irrevelant to the rendering code
4:52:20
loke
jackdaniel: I'd really like to see a configuration system for CLIM. There are a bunch of things that I need, such as hinting, font selection, etc.
4:55:10
jackdaniel
I thought it was already on quicklisp, are you sure you have a recent QL release?
4:57:32
loke
jackdaniel: In any case, I'm thinking of building one, but the question is how advanced it needs to be, and where/how to store the configuration?
5:07:39
loke
Wow... That's funny... I got a warning about unreachable code in the following snippet:
5:08:10
loke
Very weird, but it turns out it's because the SBCL optimiser decided that both parts of the last IF are the same.
5:33:47
loke
jackdaniel: For your information: I have posted the pull request, but I don't think it's quite ready for merge yet
5:36:47
loke
I think it's just more visible since freetype rendering is slower than standard rendering.
5:37:50
loke
Not entirely sure, but the fact that it's lisp vs. native doesn't matter. That's not where the difference comes from.
5:38:43
loke
The difference is that the lisp code pastes the characters to pixmaps that are then blitted to the screen. I store the symbols in a glyphmap instead and use Xrender to output the content.
5:39:34
jackdaniel
my point is that it can be faster if rendering truetype fonts doesn't cause flicker in case of the default truetype renderer
5:40:26
loke
Sorry... You lost be there. When you say “it can be faster”, what is “it” you're referring ot.
5:43:24
loke
jackdaniel: Yeah, the proper way to do it would be to not redraw everything when just a single line have changed.
5:44:15
jackdaniel
I think that only relevant region is repainted when you do output on output-recording-stream
5:45:11
loke
But the visual appearance can be improved if double-buffering is used. is there some simple way to achive this?
5:47:18
jackdaniel
afaik double buffering was buried in code at some point of time, but it was broken (according to comments from 2003, I didn't try it)
5:50:38
loke
Since Freetype support arbitrary transformation matrixes, I'd like to implement support for that.
5:51:03
loke
But I can't understand how to find the current transformation that should be applied to the text that I am drawing.