RE: Open Font License FAQ updated!

Subsetting is something that usually happens prior to a font being packaged as EOT. The same is true for WOFF, so I would consider this to be completely out of scope for both WOFF and EOT. I agree with Sylvain that there is nothing in the OpenType to EOT conversion process that alters the font (neither outlines nor hints not anything else) - the results of using the fonts extracted from EOT are identical to using the original file.

The only caveat is that 'glyf' table in fonts compressed with MTX do not always round-trip to be binary identical because MTX compression/decompression procedure optimizes font hinting whenever possible, achieving better compression efficiency (e.g. it replaces sets of individual PUSH instructions with NPUSHB or NPUSHW where applicable, etc.). This would be conceptually similar e.g. to how a "C" compiler works - the same source code compiled with different optimization level settings would generate different binary executables that produce identical results. 

Any other font data tables, including metadata contained in the 'name' table, remain intact and completely unaffected - the MTX compression is truly lossless. In essence, since the glyph design, the rendering behavior and the results of executing hint instructions would be identical for both the original TTF/OTF and the font converted from EOT format - I would agree with Sylvain that claims of EOT conversion resulting in font alterations are completely unfounded, IMHO. And when it comes to EOT-Lite/CWT - the new language of OFL FAQ claiming font alterations is plain wrong! Font data is not even touched in those cases, adding the EOT header is the only step performed in the process of conversion to EOT (e.g. as implemented by http://code.google.com/p/ttf2eot/).


Regards,
Vlad


> -----Original Message-----
> From: public-webfonts-wg-request@w3.org [mailto:public-webfonts-wg-
> request@w3.org] On Behalf Of Christopher Slye
> Sent: Tuesday, August 24, 2010 5:24 PM
> To: Sylvain Galineau
> Cc: WOFF Working Group
> Subject: Re: Open Font License FAQ updated!
> 
> Well, isn't it true that most real-world cases of EOT (i.e. WEFT-
> produced) result in some subsetting? Seems to me that EOT can be
> lossless, but often isn't -- whereas WOFF is necessarily lossless.
> 
> -C
> 
> On Aug 24, 2010, at 12:22 PM, Sylvain Galineau wrote:
> 
> >> From: public-webfonts-wg-request@w3.org [mailto:public-webfonts-wg-
> >> request@w3.org] On Behalf Of Dave Crossland
> >
> >
> >> Question: 2.3 What about other webfont formats such as
> >> EOT/EOTLite/CWT/etc.?
> >>
> >> Answer: In most cases these formats alter the original font data
> more
> >> than WOFF, and do not completely support appropriate metadata, so
> >> their use must be considered modification and RFNs may not be used.
> >
> > It's true that EOT does not have WOFF's metadata support, claiming
> that
> > EOT alters the original font data more than WOFF is puzzling. Surely,
> > prepending a header to the original file and compressing the whole
> thing
> > is less of an alteration of the font data proper than extracting each
> > table and compressing each individually.
> >
> >
> >
>

Received on Tuesday, 24 August 2010 22:43:46 UTC