Fontconfig - Nov 2005 - monospaced fonts and combining characters: fontconfig bug?

On Wed, 13 Jul 2005, Keith Packard wrote:
> Yeah, probably. Is there a compact representation for all of the
> combining characters?
The Unicode General Category of the characters can tell you that.
I can prepare a compact table if you need.
> Or should we just permit advances of 0 for any
> character? The latter would be far easier, and would follow the method
> used to identify dual-spaced fonts, while the former might avoid
> mis-identifying some broken fonts.
I prefer the latter.  Moreover, there was the discussion of
monospace fonts on the OpenType list last week.  Seems like
according to the OpenType spec, a monospace font should have all
the widths equal, no zero width allowed.  Combining characters
can get their width zeroed using OpenType tables.  There were
suggestions to change this behavior though.
> -keith
--behdad
http://behdad.org/

On Wed, 2005-07-13 at 17:12 +0200, Mike FABIAN wrote:
> Probably fontconfig should ignore all combining characters when
> checking whether a font is monospaced.
Yeah, probably. Is there a compact representation for all of the
combining characters? Or should we just permit advances of 0 for any
character? The latter would be far easier, and would follow the method
used to identify dual-spaced fonts, while the former might avoid
mis-identifying some broken fonts.

-keith

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url :
http://lists.freedesktop.org/archives/fontconfig/attachments/20050713/43a00343/attachment.pgp

On Wed, 2005-07-13 at 12:45 -0400, Behdad Esfahbod wrote:
> I prefer the latter.  Moreover, there was the discussion of
> monospace fonts on the OpenType list last week.  Seems like
> according to the OpenType spec, a monospace font should have all
> the widths equal, no zero width allowed. 
Sure, but these aren''t OpenType fonts, they''re just TrueType,
so there
isn''t any way to set the glyphs correctly.
>  Combining characters
> can get their width zeroed using OpenType tables.  There were
> suggestions to change this behavior though.
I expect applications wanting monospaced fonts will mis-display these
glyphs quite badly, and I''m not quite sure what to do about that; I
know
my text editor doesn''t expect glyphs in a monospaced font to overlap...

-keith

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url :
http://lists.freedesktop.org/archives/fontconfig/attachments/20050713/441a4fd6/attachment.pgp

On Wed, Jul 13, 2005 at 10:43:18AM -0700, Keith Packard
wrote:
> On Wed, 2005-07-13 at 12:45 -0400, Behdad Esfahbod wrote:
> 
> > I prefer the latter.  Moreover, there was the discussion of
> > monospace fonts on the OpenType list last week.  Seems like
> > according to the OpenType spec, a monospace font should have all
> > the widths equal, no zero width allowed. 
> 
> Sure, but these aren''t OpenType fonts, they''re just
TrueType, so there
> isn''t any way to set the glyphs correctly.
If I am not mistaken, the discussion on the OpenType list should be
inspired by the author of FontForge, who got asked about exactly this
issue for TrueType fonts on FontForge list.

The thread is:
   
http://sourceforge.net/mailarchive/forum.php?thread_id=7578042&forum_id=39351
They were arguing whether a font of glyphs with only a certain width or
zero width should be labelled as monospace by FontForge or not.  The
author of FontForge insists that it should not, and monospace font means
all glyphs have the same width.

I don''t know much about fonts, just to point that thread to the list,
so
that people with more knowledge can evaluate the opinions there.

Also, a sincere thank you to the fontconfig developers.

Ming, a happy Chinese Linux user
2005.07.14

Keith Packard <keithp@keithp.com> ????????:
> Yeah, probably. Is there a compact representation for all of the
> combining characters? Or should we just permit advances of 0 for any
> character? The latter would be far easier, and would follow the method
> used to identify dual-spaced fonts, while the former might avoid
> mis-identifying some broken fonts.
Permitting advances of 0 for any character sounds like a good idea
which is probably good enough for the moment.

If it turns out that there are really some broken fonts which are
mis-identified because of this, we could still add a list of combining
characters and permit advances of 0 only for characters in that list.

-- 
Mike FABIAN   <mfabian@suse.de>   http://www.suse.de/~mfabian
?????????????

Fontconfig fails to detect that the "FreeMono" fonts are monospaced,
it doesn''t set spacing=100:

mfabian@magellan:~$ fc-list FreeMono family style spacing 
FreeMono:style=BoldOblique,Negreta cursiva,tu?n? kurz?va,fed kursiv,Fett
Kursiv,?????? ??????,Negrita Cursiva,Lihavoitu Kursivoi,Gras Italique,F?lk?v?r
d?lt,Grassetto Corsivo,Vet Cursief,Halvfet Kursiv,Pogrubiona kursywa,Negrito
It?lico,?????????? ??????,Tu?n? kurz?va,Fet Kursiv,Kal?n ?talik,polkrepko
le?e?e,Lodi etzana
FreeMono:style=Bold,Negreta,tu?n?,fed,Fett,??????,Negrita,Lihavoitu,Gras,F?lk?v?r,Grassetto,Vet,Halvfet,Pogrubiony,Negrito,??????????,Fet,Kal?n,polkrepko,???m,Lodia
FreeMono:style=Medium,Normal,oby?ejn?,Standard,????????,Normaali,Norm?l,Normale,Standaard,Normalny,???????,Norm?lne,navadno,th???ng,Arrunta
FreeMono:style=Oblique,Cursiva,kurz?va,kursiv,??????,Kursivoitu,Italique,D?lt,Corsivo,Cursief,Kursywa,It?lico,??????,?talik,le?e?e,nghi?ng,Etzana
mfabian@magellan:~$

When checking why this happens, I found that the advance widths for
almost all glyphs in the "FreeMono" fonts are 600, *except* for U+0300
- U+0344 where the advance width is 0.

But all the characters in the range U+0300 - U+0344 are combining
characters and as I currently understand it, it is correct that the
advance width is 0 for combining characters.  It is like that in most
other fonts as well, for example in "Arial Unicode MS" (not
monospaced). "Courier New" is an exeption, it has the same advance
width for combining characters as for regular characters, therefore
fontconfig recognizes "Courier New" as monospaced.

Probably fontconfig should ignore all combining characters when
checking whether a font is monospaced.

Is that correct?

See also:

http://lists.gnu.org/archive/html/freefont-bugs/2003-08/msg00000.html
http://lists.gnu.org/archive/html/freefont-bugs/2003-09/msg00001.html
http://lists.gnu.org/archive/html/freefont-bugs/2003-09/msg00002.html


-- 
Mike FABIAN   <mfabian@suse.de>   http://www.suse.de/~mfabian
?????????????