tynet-lichat/shirakumo - IRC Chatlog
Search
9:40:19
shinmera
selwyn: I mean, if you wanted things to be as fast as they could be you'd probably be a russian working for intel. :)
10:32:44
shinmera
The people writing the high performance computing libraries for Intel and so forth in ASM are predominantly russian
10:32:49
selwyn
i remember some japanese wizard who beats blas implementations on speed by handcrafting assembly
11:43:01
Colleen
Indeed. Worse that stassats and I e.g. agreed that one part of SBCL bootstrapping should not work, yet it does work in upstream (but not in my fork, hence my finding it).
11:43:01
hayley
Indeed. Worse that stassats and I e.g. agreed that one part of SBCL bootstrapping should not work, yet it does work in upstream (but not in my fork, hence my finding it).
12:35:40
shinmera
|3b|: I'm reaching the point where characters being a plist in the fontspec are a problem. (mostly because it can't type derive the values.
12:38:30
shinmera
Also, would it be OK to change all the values to be single-floats, or do you want to keep the precise ratios?
13:04:56
shinmera
I'm also a bit confused: in the write-bmfont-* functions you use GETF to access the font fields, but the actual font object is a class, no?