- From: Chris Lilley <chris@w3.org>
- Date: Wed, 25 Aug 2010 09:39:29 +0200
- To: Christopher Slye <cslye@adobe.com>
- CC: Sylvain Galineau <sylvaing@microsoft.com>, WOFF Working Group <public-webfonts-wg@w3.org>
On Tuesday, August 24, 2010, 11:23:36 PM, Christopher wrote: CS> Well, isn't it true that most real-world cases of EOT (i.e. CS> WEFT-produced) result in some subsetting? Seems to me that EOT can CS> be lossless, but often isn't -- whereas WOFF is necessarily lossless. And this confusion between what a given tool does and what a spec says, is why I really want the WOFF spec to describe a processing model where any subsetting, name table rewriting, excess padding removal, checksum recomputation is seen as a preprocessing step. The result of that step is the 'origin' font. Converting that to WOFF and decoding back yields a 'result' font which is bit for bit identical to the 'origin' font, and this is the definition of the 'losslessness' of WOFF. There may be a postprocessing step as well (like the sanitization that the Chrome browser performs). (I have no particular attachment to the names I used here, they could easily be changed to better ones). A diagram would be handy here also. -- Chris Lilley Technical Director, Interaction Domain W3C Graphics Activity Lead, Fonts Activity Lead Co-Chair, W3C Hypertext CG Member, CSS, WebFonts, SVG Working Groups
Received on Wednesday, 25 August 2010 07:38:48 UTC