[XeTeX] Problem with Unicode 01F01B

Steve White stevan.white at gmail.com
Wed Dec 19 14:22:49 CET 2012


Hi Khaled,

Pander just alerted me about this thread, and I looked into it.

RE fontlint test
================
I don't see anything like your output, either on the last official
release of GNU FreeFont (20120503), or its latest SVN, either with my
distro release of fontlint (Ubuntu fontforge, from 22-Feb-2011-ML), or
with my own build of fontlint, from sources earlier this year.

See below an example of my output of fontlint runs on FreeSerif.

** Please report the version of FreeFont, and the version of
fontforge, you used to get these results.

** If you used any special run flags for fontlint, please report those too.

RE hinting
==========

First, the GNU FreeFont sources do not themselves specify hinting.
Hinting is added by fontforge as part of the build process.  I would
have to characterize issues with hinting as problems with the latter
software.

Second, I don't recognize the problem
   "circle3mahjong glyph has hstemhm hint command with 49 arguments,"
where can we see this?

many problems
=============
You remark that FreeFont "many problems": you have made such remarks
elsewhere at various times.

** We remain eager to see your problem reports.

FreeFont is a big and complex project.  It certainly does have many
problems, as does any big complex project.  We rely on the teamwork of
the community to improve the software.

Mahjong Tiles
=============

The glyphs in question, in the Mahjong Tiles range U+1f000 - U+1f02b
are indeed very complicated.  I personally spent a lot of time cleaning
the glyphs up, and arranging for simpler representation.  Still some of
them contain the largest number of nodes of any glyph in the whole font.

I wouldn't be at all surprised that they would cause some software to trip up.

It appears, however, we still don't know what in the font causes the trip.

=========================================================
Typical output of fontlint:

# freefont-20120503] /usr/bin/fontlint FreeSerif.ttf
Copyright (c) 2000-2011 by George Williams.
 Executable based on sources from 13:48 GMT 22-Feb-2011-ML.
 Library based on sources from 13:48 GMT 22-Feb-2011.
This font contains both a 'kern' table and a 'GPOS' table.
  The 'kern' table will only be read if there is no 'kern' feature in 'GPOS'.
Validation FreeSerif ...Failed
  A glyph uses at least one, but not all, anchor classes in a subtable
   (I'm not absolutely sure this is an error)

=========================================================
RE the typical output:

The fontlist test fails due to the presence of two different kerning
tables, but I don't regard this as a bug, and I deliberately include
both versions ot the tables for good reasons.

It also warns about anchor classes, but qualifies its warning.
I can't imagine why it complains at all.

In any case, I don't think these have anything to do with the issue at hand.

Cheers!

On Tue, Dec 18, 2012 at 3:36 PM, Khaled Hosny <khaledhosny at eglug.org> wrote:
> On Tue, Dec 18, 2012 at 12:37:14PM +0100, Pander wrote:
>> On 2012-12-18 12:18, Khaled Hosny wrote:
>> > On Tue, Dec 18, 2012 at 12:11:04PM +0100, Pander wrote:
>> >> On 2012-12-17 19:04, Khaled Hosny wrote:
>> >>> On Mon, Dec 17, 2012 at 06:47:19PM +0100, Pander wrote:
>> >>>> Hi all,
>> >>>>
>> >>>> I am getting errors with XeLaTeX for Unicodes 01f01b up to 01f030.
>> >>>> Attached is a test that fails and results in:
>> >>>>
>> >>>> ** ERROR ** Type2 Charstring Parser: Parsing charstring failed:
>> >>>> (status=-2, stack=48)
>> >>>>
>> >>>> Output file removed.
>> >>>>  )
>> >>>> Error 256 (driver return code) generating output;
>> >>>> file test.pdf may not be valid.
>> >>>>
>> >>>> Strange thing is that 01f01a is processed correctly and beyond 01f030
>> >>>> also. I have asked the maintainer of GNU FreeFonts to test this but he
>> >>>> cannot reproduce this problem. Can any of you?
>> >>>>
>> >>>> I am using latest TeX Live and latest GNU FreeFonts.
>> >>>
>> >>> Runs fine here, but this sounds like a font-related xdvipdfmx error.
>> >>>
>> >>> Try running:
>> >>>
>> >>>   xelatex --no-pdf test
>> >>
>> >> xelatex --no-pdf test
>> >> This is XeTeX, Version 3.1415926-2.4-0.9998 (TeX Live 2012)
>> >>  restricted \write18 enabled.
>> >> entering extended mode
>> >> (./test.tex
>> >> LaTeX2e <2011/06/27>
>> >> ...
>> >> [1] (./test.aux) )
>> >> Output written on test.xdv (1 page, 660 bytes).
>> >> Transcript written on test.log.
>> >>
>> >>>   xdvipdfmx -v test
>> >>
>> >> xdvipdfmx -v test
>> >> <FONTMAP:pdftex.map><FONTMAP:kanjix.map><FONTMAP:ckx.map>
>> >> ** ERROR ** Could not open specified DVI (or XDV) file: test
>> >>
>> >> Output file removed.
>> >
>> > My bad, use
>> >   xdvipdfmx -v test.xdv
>>
>> "/usr/local/texlive/2012/texmf-dist/fonts/opentype/public/gnu-freefont/FreeSerif.otf"...
>> ** ERROR ** Type2 Charstring Parser: Parsing charstring failed:
>> (status=-2, stack=48)
>
> The FreeSerif.otf font has several issues, in this particular test file
> the circle3mahjong glyph has hstemhm hint command with 49 arguments,
> more than the 48 limit specified by Type 2 Charstring Format technote.
>
> If you run fontlint (from FontForge) on the font you get many warnings,
> like:
>
> Stack got too big in circle3mahjong
> Stack got too big in circle5mahjong
> Stack got too big in circle6mahjong
> Stack got too big in circle7mahjong
> Stack got too big in circle8mahjong
> Stack got too big in circle9mahjong
> Stack got too big in chrysantemummahjong
> Stack got too big in mj_circle7_hi
> In FreeSerif, in glyph circle3mahjong, 'CFF ' advance width (0) and 'hmtx' width (795) do not match.
>
>> > (note the extension, sorry I didn't realize that TL version does not
>> > auto-look for .xdv files).
>>
>> Feature request?
>
> The SVN version do already.
>
> Regards,
> Khaled
>
>
> --------------------------------------------------
> Subscriptions, Archive, and List information, etc.:
>   http://tug.org/mailman/listinfo/xetex


More information about the XeTeX mailing list