On Mon, 2006-02-27 at 14:13 -0700, Greg Haerr wrote: > What I'm saying is that the font buffers could be shared > between all characters of all fonts. By keeping track of the > font number as well as the character, all displayed characters > would be cached from the same area, subject to LRU loading. > The entire font cache could be made larger (or smaller) as > desired, without having to pre-allocate five times the memory > for five fonts, as is required now.
OK, now I think I understand what you're saying. The unified cache would still be static though, right? Instead of five 10kb (or 45kb or whatever) buffers we'd have a single (probably larger, somewhere around 50kb?) buffer used for all fonts, with everything going through the font cache & LRU code rather than simple static loading. That does seem like a cleaner, probably faster way of doing things (faster on the loading side, hopefully reasonably fast wrt individual glyph access). bk
