Fontforge join points6/11/2023 Kevin: FontForge may be fragile when dealing with this font, but in fairness it *did* produce a big bunch of relevant warnings! I will send the relevant line of my patch upstream for the sake of robustness. I don't understand how this came about, but they will need correcting too. There are several glyphs with the same name in this font as well, and a few that are mapped to the same Unicode point. ![]() (Works for me with upstream CVS head, haven't tried Rawhide). Do same as above (there are two of them here). Remove dots that are disconnected from main glyphĤ. Go to glyph "cent" (the second one with this name, U+0BBF). For both serafettin-cartoon.sfd and serafettin-cartoon-italic.sfd:ģ. Long story short, run the base font through FontForge and clean up the glyphs. These don't have any effect on the drawn outline, but they mess with FontForge's ability to determine the direction of each part of the contour when stroking the lines in order to expand or contract them. If you go into FontForge and validate, you'll find a lot of glyphs that don't appear to be self-intersecting, but if you look at the SFD source, it will be clear that there are a lot of points on top of one another. There are several interesting lines here, but the most important for this bug appears to be "self intersecting glyph". References nested more deeply than specified in 'maxp' ![]() More references in a glyph than specified in 'maxp' I've had another good look at Serafettin and I can see why FontForge has such a hard time with it. Unless someone else get there first, I'll try the current CVS version of FontForge tonight, but I wouldn't be surprised if this crash has already been fixed in CVS. Splinestroke.c is now at version 1.62 in CVS. (Built direct from the tarball, as I couldn't find this RPM.)Ī new algorithm for stroking splines had been introduced between the two releases (at splinestroke.c version 1.48), and the CVS log shows quite a few oopses being fixed since that time. I can confirm oget's finding of a crash in this version. Splinestroke.c was at version 1.46 at that time.įontForge 20100501 included splinestroke.c version 1.56. When using 'Merge' in FontForge, the resulting merged curve is quite inaccurate in many cases. ![]() The boldening algorithm throws a lot of warnings, but no crashes. I can confirm that Serafettin builds OK with FontForge 20090923 (i.e. Nevertheless, from the backtrace we see that the crash occurred in HideStrokePointsCircle(), which is in splinestroke.c. Firstly, was this a scratch build? This bug has been reported against a version that doesn't appear in the updates for F13.
0 Comments
Leave a Reply. |