abrt detected a crash. Attached file: backtrace cmdline: /usr/bin/fontforge -lang=ff /usr/bin/fontlint usr/share/poker3d/data/neurpoli.ttf component: fontforge executable: /usr/bin/fontforge kernel: 2.6.31.5-122.fc12.x86_64 package: fontforge-20090923-1.fc12 rating: 3 reason: Process was terminated by signal 11
Created attachment 369093 [details] File: backtrace
reported upstream.
I'm not sure if this one was fixed or not. Do you see it with current rawhide fontforge?
This still segfaults with the scratch build you made on 2010-07-04: http://koji.fedoraproject.org/koji/taskinfo?taskID=2294313 This is the build with sources from 2010-05-01. It also crashes with current CVS, sources from 2010-06-25.
Created attachment 431233 [details] Smallest subset of original font that produces crash I have cut the original font down to the smallest subset that reproduces the original crash, and attached it here. Validation crashes on the ellipsis character, which is made by reference to two other characters, full stop and space. The references to space are pointless, as it has no outlines. I am about to send a bug report to the upstream mailing list with this information.
Thanks Paul. :) Hopefully upstream will find a fix.
Created attachment 431739 [details] Correct LayerUnAllSplines() when refs have nil splinesets When validation is performed, the set of splines for each character, including those used by reference (diacritics, for example), are "flattened" to one set by LayerAllSplines(). LayerUnAllSplines() is supposed to perform the reverse transformation, but fails if any references have nil splinesets. This patch applies to 20090923, for the benefit of F13 and rawhide. It will be supplied upstream for CVS head with more information this evening. Local testing: [paul@palant x86_64]$ fontlint /usr/share/poker3d/data/neurpoli.ttf Copyright (c) 2000-2009 by George Williams. Executable based on sources from 23:48 GMT 23-Sep-2009. Library based on sources from 17:32 GMT 14-Sep-2009. The following table(s) in the font have been ignored by FontForge Ignoring 'LTSH' linear threshold table Ignoring 'VDMX' vertical device metrics table Ignoring 'hdmx' horizontal device metrics table The glyph named periodcentered is mapped to U+2219. But its name indicates it should be mapped to U+00B7. The glyph named Delta is mapped to U+2206. But its name indicates it should be mapped to U+0394. Validation Neuropolitical ...Failed Self Intersecting Glyph Missing Points at Extrema
I can go ahead and add this patch in rawhide... or did we need to backport it to the f12/f13 versions? Typically we don't upgrade fontforge in a stable release to avoid any regressions with fonts already built with it. Thoughts?
I'd patch it in rawhide. The chances of this occurring with any other font are extremely slim (and I ran fontlint on 30000 of them yesterday). If bug 615347 is accepted, this font won't be in Fedora for long anyway ;-)
ok. Built for rawhide: http://koji.fedoraproject.org/koji/taskinfo?taskID=2324995 Shall we close this bug now?
(In reply to comment #10) > > Shall we close this bug now? I vote yes.