Log of #mupdf at irc.freenode.net.

Search:
 <<<Back 1 day (to 2017/05/24)20170525 
tor8 Robin_Watts: ping.10:18.06 
Robin_Watts pong10:43.38 
tor8 two commits on tor/master10:44.12 
  I wanted to ask where you got the default replacement font->bbox values for invalid font bboxes from10:44.29 
  Make that 3 commits. Ignore the top 3.10:44.49 
Robin_Watts tor8: No idea. I'd hope the commit messages in question have pointers.10:47.43 
tor8 it does not, and the numbers were cryptic :)10:48.00 
  my commit fixes the bug mentioned in the "Text bounds - 65075" email to support10:48.42 
  by unioning the type3 computed bboxes with the font bbox if it's invalid10:49.11 
  the file in the bug has a crazy stupid font matrix and font size10:49.28 
Robin_Watts tor8: So... now, if we run a page to a display list, we have no way of saying not to put images into the list ?10:49.38 
tor8 the font size is 0.24 pixels, but the glyphs are about 10x bigger than the unit10:50.08 
  Robin_Watts: we do not. nor did we before... the FZ_IGNORE_IMAGES hint never got propagated to the interpreter.10:50.34 
Robin_Watts tor8: oh, right. the unit square would have been 0,0, 1,1.10:51.54 
  We extended that by 1 either side, hence -1,-1, 2,210:52.06 
  Urgh, that is a poor commit message, yes.10:53.10 
  I blame the reviewer :)10:53.14 
  Can we update the commit message so we don't make the same mistake again please?10:54.29 
  Either list a bug number, or a command line/file that shows the problem.10:54.44 
  With that commit message tweak, all 3 lgtm.10:55.33 
tor8 I'm just not sure what to make of it, I'd be happy to keep the default at -1,-1,2,2 (and just let the glyph bboxes extend it)10:56.03 
  I suspect we should be good with it being 0,0,0,0 now, since type3 fonts will recompute it from the glyphs10:56.28 
  and we generally should have the proper bbox form type1/truetype font files themselves10:56.43 
  oh... now I suspect I know why some files have really tall text bboxes.11:05.58 
  Robin_Watts: updated bbox commit on tor/master with some different behaviour and more comments11:14.39 
Robin_Watts tor8: Update the commit to mention some filenames in the commit message, and then, assuming it passes the cluster, lgtm.11:23.00 
  The important thing is that if we hit problems in this area in the past we get to look back at this commit and see the files that made us make these changes.11:23.31 
  tor8: mention the bug number!12:14.26 
ej help13:41.44 
 Forward 1 day (to 2017/05/26)>>> 
ghostscript.com #ghostscript
Search: