Patrick Lam <plam@xxxxxxx> さんは書きました: > I've committed this patch. I hope it's alright; I couldn't see anything > wrong with it. If I understand correctly, the crux of the correctness > issue is the change of > > score += v2_best_score; > > to > > v = score; > > (which would be score = v2_best_score in the old code). > > and the rest is a performance optimization. Is that right? Could > someone verify that this change alone fixes the bug? I cannot find exactly the two lines you mention above. There is a similar part in Zhe Su's patch though which I have attached here. Applying *only* this part does *not* solve the problem, actually it makes it worse: mfabian@magellan:/var/tmp/abuild/x86_64$ LC_ALL=zh_CN.UTF-8 fc-match 方正楷体 ethio24f-uni.bdf: "ethio24f-uni" "Regular" mfabian@magellan:/var/tmp/abuild/x86_64$ LC_ALL=zh_CN.UTF-8 fc-match FZSongTi ethio24f-uni.bdf: "ethio24f-uni" "Regular" mfabian@magellan:/var/tmp/abuild/x86_64$ LANG=ja_JP.UTF-8 fc-match sans ethio24f-uni.bdf: "ethio24f-uni" "Regular" Zhe Su's complete patch does really solve the problem though.
diff -ru fontconfig-2.3.93.20060105.orig/src/fcmatch.c fontconfig-2.3.93.20060105/src/fcmatch.c --- fontconfig-2.3.93.20060105.orig/src/fcmatch.c 2005-11-30 11:03:18.000000000 +0100 +++ fontconfig-2.3.93.20060105/src/fcmatch.c 2006-01-09 12:52:32.000000000 +0100 @@ -640,10 +640,9 @@ /* I'm actually kind of surprised that * this isn't v + 100 * v1_offset. -PL */ v = v * 100 + v1_offset; - if (v < v2_best_score) - v2_best_score = v; + if (v < score) + score = v; } - score += v2_best_score; } }
-- Mike FABIAN <mfabian@xxxxxxx> http://www.suse.de/~mfabian 睡眠不足はいい仕事の敵だ。
_______________________________________________ Fontconfig mailing list Fontconfig@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/fontconfig
