RE: Web font test cases: file linking vs. data URI embedding, CSS font stack subsets

The difference between "splitting" and "subsetting" is that by
subsetting a font you select and keep all the data that you do need [to
layout and render a text on a page] but eliminate the font data that you
don't need - this makes the font file smaller but 100% functional and
usable.

If you do need to have a complete font but decide to split it into few
separate font subsets (e.g. for the purposes of obfuscation, or to speed
up the font download), you will likely end up with a 'broken' font
unless you pay careful attention to what glyph pairs have kerning
information, how a font is hinted and whether there are dependencies
between glyphs (e.g. when a glyph can be described as composite glyph
using few simple glyphs), what glyph combinations are used to create
ligatures, etc. This may be doable for a limited set of fonts supporting
simple scripts with no advanced features, but I agree with John that
font splitting as a general-purpose mechanism is a non-starter.

Regards,
Vlad


> -----Original Message-----
> From: www-font-request@w3.org [mailto:www-font-request@w3.org] On
> Behalf Of Richard Fink
> Sent: Tuesday, October 13, 2009 10:26 AM
> To: 'John Hudson'; 'Ben Weiner'
> Cc: 'www-font'; 'John Daggett'
> Subject: RE: Web font test cases: file linking vs. data URI embedding,
> CSS font stack subsets
> 
> Tuesday, October 13, 2009 3:29 AM John Hudson <tiro@tiro.com>:
> 
> >splitting fonts into multiple files is fraught with peril.
> Please explain what you mean by splitting, John. Splitting just for
the
> sake
> of obfuscation or a ham-fisted Fontlab fumble that I might attempt?
> (Frightening.)
> But smart, knowledgeable sub-setting to keep file sizes small and
> on-purpose, would seem to be a valuable service and/or product to
> provide in
> light of the file sizes involved. Indeed, this seemed to be one of the
> promises of Microsoft's WEFT tool. Is it not truly doable?
> 
> >I think font splitting is a bit of a
> >non-starter, at least as a general purpose mechanism.
> Once again, "splitting" means...?
> 
> Regards,
> Rich
> 
> -----Original Message-----
> From: www-font-request@w3.org [mailto:www-font-request@w3.org] On
> Behalf Of
> John Hudson
> Sent: Tuesday, October 13, 2009 3:29 AM
> To: Ben Weiner
> Cc: www-font; John Daggett
> Subject: Re: Web font test cases: file linking vs. data URI embedding,
> CSS
> font stack subsets
> 
> Ben Weiner wrote:
> 
> > Would joined/substituted glyphs fail in a similar way if, say, an
> Arabic
> > font was split up into subsets? Our discussion is very Latin-centric
> ;-)
> 
> Yes. OpenType Layout is processed in glyph runs, and a change of font
> breaks a run.
> 
> As John D says, splitting fonts into multiple files is fraught with
> peril. It can be done cleverly for simple scripts that don't need
> complex layout, e.g. by parsing the kern data and avoiding putting
> glyphs with a kerning relationship into separate fonts, but then the
> whole value of font splitting becomes dependent on how extensively
> hinted a font is. On the whole, I think font splitting is a bit of a
> non-starter, at least as a general purpose mechanism.
> 
> John H.
>

Received on Tuesday, 13 October 2009 16:42:48 UTC