freenode/#clim - IRC Chatlog
Search
13:41:09
froggey
I think accessors should be faster, but there's not much in it if slot-value is called with a known slot name (slot-value foo 'slot). with-slots is built on slot-value and has the same performance
13:42:50
froggey
most of the perf issues are probably mezzano-related. the compiler isn't so great in places and method dispatch can get really slow when there are methods with eql specializers and specializers on multiple arguments
13:50:53
jackdaniel
fittestbits: clx-fb sends an image via wire-protocol after drawing it in the memory – how does it work on mezzano port?
13:51:31
jackdaniel
because there is an optimization possible if you can take a framebuffer to draw on it directly (no need to copy image and blit it after drawing)
13:53:50
jackdaniel
so is it possible to just pass the framebuffer to the renderer to draw directly?
13:54:01
fittestbits
I'll look into that. Also, I think there are places where the common case can be optimized while still handling the general case in some of the clipping routines.
13:55:04
jackdaniel
go for it, I have (shamefully) a very unproductive period with my coding lately, I've failed to find the cause yet (maybe I'm just a very lazy person though)
13:55:43
fittestbits
It should be possible to pass the framebuffer. Mezzano may even use double buffering at the framebuffer level, it's been a while since I looked at it.
13:56:25
jackdaniel
alright. in that case you should see a noticeable speedup when you do that (I think)