[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [Pkg-fonts-devel] GNU FreeFont 20120503 released

Hi Christian,

On Sat, Jun 9, 2012 at 8:09 AM, Christian PERRIER <bubulle@debian.org> wrote:
> (cc'ing you as I don't remember if you're subscribed to our mailing list)
> Quoting Steve White (stevan.white@googlemail.com):
>> The deal is, due to a disagreement between people working on
>> FontForge, regarding the interpretation of
>> the standards, a flag is turned off.  This used to only affect Windows
>> applications, but now it affects Linux too.
>> I built the font with a modified version of FontForge, which (I
>> believe correctly) sets the flag.
>> I submitted the patch to FontForge, and expected it would be implemented soon.
>> But others have objected (for reasons whose explanation still escapes
>> me).   So the patch isn't in.
> Wouldn't there be a possibility to set this flag optionnally in FF,
> through a command line switch?
> This way, people could use the same FF and decide to use the flag or
> not use it.
> I don't know anything from FF code, but that could be an interesting
> option if it was possible.
I would go for this idea.

However at this point, I don't think it really has anything to do with
what works best.
There were evidently some flame wars, and people got their backs up
against a wall.
(I wasn't aware of this when I submitted the patch.)

The way it works now, one sets the OS/2 monospace attribute, and FF does
a loop through all the glyphs, checking that most of the glyphs (with
a few exceptions)
are of the same width, and if so, it sets the flag, otherwise it un-sets it.

However, the TrueType Manual clearly states that zero-width glyphs are allowed
in a monospace font (the OpenType spec even recommends them), and of
course there
are all those combining marks, and explicitly zero-width characters
etc in Unicode.
My patch simply changed the loop to set the flag as stated in the
spec: all the same
positive width or zero width.

The argument against this I cannot understand.  First that the spec
was unclear (in what
respect, I could not understand.  Maybe because it didn't say what
they wanted it to say.)
 and then that such a setting would break some Windows apps (but no
examples were
provided -- just a referece to a mailing list...I followed this lead,
found a long conversation,
but... none  of what I found there indicated a failure caused by
setting this flag.  Confusion.)

Now, I tried it in Windows (three different versions) on dozens of
apps.  I saw none of the
predicted problems.  All I saw was, without the flag, a font was not
recognized as
monospace; with it, it appears on lists of monospace fonts.

A separate checkbox for the flag would be OK for me.  I don't think
the OS/2 setting
and the flag are tied together by any specification.  But as I said,
it doesn't seem to be
about what would work best any more.

So... I'm considering other options, such as setting the flag
*outside* of FontForge in
the build process.  It's a cludge but it would work.  Just, it'll take
a day or two to write it.


Reply to: