> Currently, I can't debug the fpgm/prep tables with FontForge because > it crashes. Rogier, maybe this information gives you further hints. > It seems that we have wrong assumptions on the storage area, and how > RS/WS interact with the CVT.
George provided a quick patch, so I was able to debug it. I have found the problem, which is something completely different: Function #84 (defined in the <fpgm> table) in font verdanab.ttf (version 2.40) is called as the last subroutine in the <prep> table. Within the function, the GETINFO instruction is called to find out whether we do grayscaling. If yes, storage[2] is set to 1 and to 0 otherwise. The <gasp> table indicates that you must not use gray scaling in the range 9-16, so everything is fine. However, verdanab.ttf exhibited two bugs in FreeType, which I've now fixed: . `tt_loader_init' never reset `exec->grayscale' properly because a wrong constant has been used in a comparison. . A change from mono to grayscale rendering (and vice versa) requires a re-execution of `tt_face_load_prep'. Werner _______________________________________________ Freetype mailing list Freetype@nongnu.org http://lists.nongnu.org/mailman/listinfo/freetype