http://bonsai.mozilla.org/cvsblame.cgi?file=mozilla/gfx/cairo/cairo/src/cairo-cff-subset.c&rev=1.1&mark=167,171#168 [167] } else if (i >= -1131 && i <= -108) { 168 i = -i - 108; 169 vladimir 1.1 *p++ = (i >> 8)+ 251; 170 *p++ = i & 0xff; [171] } else if (i >= -1131 && i <= -108) { if the condition at [167] is false the condition at [171] is also false. if the condition at [167] is true the [171] is not reached because of else. in addition this function seems to do uselesss operation on |i|. 172 *p++ = 28;

What should a patch do here? Just remove the "else if" at line 171?

(In reply to comment #1) > What should a patch do here? Just remove the "else if" at line 171? > not quite sure, can't get the context. it is possible that the condition must be different and the block present.

I have fixed this bug in cairo 1.5.1 http://gitweb.freedesktop.org/?p=cairo;a=commit;h=f52aa4c13e91339e575ca2c52c9e3a1f4d95b106 Line 171 should be } else if (i >= -32768 && i <= 32767) { The effect of this bug is that some integers are encoded with five bytes instead of three resulting in a slightly larger subsetted font.

Trivial fix based on <http://gitweb.freedesktop.org/?p=cairo;a=blobdiff;h=55e0f1575aa9eb79a32ddd22c26722422db5c536;hp=2adc709c0bdef6e91ec9efe0234b5ba6b17122ce;hb=f52aa4c13e91339e575ca2c52c9e3a1f4d95b106;f=src/cairo-cff-subset.c>.

Near as I can tell from the clear-as-mud approval rules, you need approval1.9+ before you're checkin-needed.

vlad: can you approve this/push it upstream?

(In reply to comment #6) > vlad: can you approve this/push it upstream? The patch is my patch from upstream. As the bug doesn't actually break anything (the effect is about a 1% increase in embedded font size in PDFs) I would suggest waiting for the next cairo upgrade to pull in the fix.

(In reply to comment #7) > As the bug doesn't actually break anything (the effect is about a 1% increase > in embedded font size in PDFs) I would suggest waiting for the next cairo > upgrade to pull in the fix. So, should we close this bug?

Yes, fixed by bug 395449.