better fix
*** empty log message ***
*** empty log message ***
cleanup
use an compressed format for glyph quad vectors. this reduces the size of e.g. the opengl info from 200k to 20k, and typical sizes for labels are now <<100 bytes. the cost is full immediate mode submission of coordinates (cpu), but glInterleavedArrays is likely not much faster and modern cpus probably prefer more decoding over more cache misses.
use rendercache to speed up text reendering and hopefully also save memory
fix pango trapezoid rendering
initial rewrite of pangorenderer => rendercache
need more flushes
*** empty log message ***
minor fixes and improvements
bugs\! bugs\! bugs\!
There is of course no way to clear glyph caches in stupid pango, so we have to add a generation counter to each and every glyph structure and check it on every access.
small fixed, ported to the language called microsoft c, which superficially looks like ISO-C and is similar enough so you can write programs in their common subset.
Get rid of cairo completely (yay!) and of ft2 factually (still need the library as it included pangofc), by introducing a custom pango opengl renderer. Text rendering now no longer requires the distinction between rgba and grayscale modes, requires much less texture space and memory, and is faster on accelerated hardware (and possibly with software rendering, too). All at the cost of only 1200 lines or so.
This form allows you to request diffs between any two revisions of this file. For each of the two "sides" of the diff, select a symbolic revision name using the selection box, or choose 'Use Text Field' and enter a numeric revision.