- From: Joshua RANDALL FTRD/DIH/BOS <joshua.randall@francetelecom.com>
- Date: Tue, 25 Apr 2006 19:07:42 -0400
- To: Håkon Wium Lie <howcome@opera.com>
- Cc: Joshua RANDALL FTRD/DIH/BOS <joshua.randall@francetelecom.com>, David Woolley <david@djwhome.demon.co.uk>, www-style@w3.org
> Now that descigners are using > advanced CSS, I think the time is right try getting webfonts started > again. And simple downloads of existing, currently available, zipped > truetype files is a pragmatic way to start. I agree that it would definitely be good to reincorporate webfonts into current CSS work, but I'm not sure how a standard supporting webfonts would specify required font type support for conformance purposes. While truetype is certainly an industry standard, is it standardized by any actual standards body in a public document that could be referenced by a W3C recommendation? Perhaps SVG fonts could be considered as an alternative, not necessarily instead of truetype, but as a format that could be required for baseline conformance in addition to other formats that may be supported? Major points in favor of using the truetype format for webfonts are that fonts are widely available, that some form of truetype support is already present on most desktop operating systems, and that hinting can be used to provide high-quality rendering at arbitrary pixel sizes. However, one of the goals of the style effort, moving forward, should be to make it possible to render pages on a much wider variety of devices; including PDAs, mobile phones, and set-top boxes. In many cases these devices do not contain built-in support for truetype, so it would be up to user-agent manufacturers to provide such support. Since these devices often have a very limited set of built-in fonts (sometimes only one or two faces), they arguably have the most to gain from webfonts work. It would be unfortunate to embark on a new standardization effort for webfonts that alienates those users who would benefit the most. As for hinting support, this portion of truetype is potentially encumbered by patents and therefore support for hinting may not be able to be provided by an unlicensed system (such as open source/free software). Given this, I don't see how a W3C recommendation could _require_ a full truetype rendering system. Perhaps a recommendation could require a level of truetype support that does not include hinting, but that nullifies one of the major benefits of the format. It seems to me that having a webfont system that only supported truetype would be much like the early days of HTML, when the primary image format supported was GIF. Years later, the W3C finally created the PNG recommendation, I presume at least in part so that the other W3C recommendations could refer to a format that was an open standard and unencumbered by patents. Luckily, the situation today with webfonts is not completely analogous to that of the early web and GIF images. In the absence of an open raster format such as PNG, it made sense to support GIF -- but we _do_ have an open standard for specifying fonts (albeit rather underutilized) in SVG. I can't imagine how webfonts could be done without at least including support for fonts defined as SVG fonts. They could provide a baseline conformance level that all visual user- agents could be expected to implement, are defined in an open standard that can easily be referenced, can potentially be included directly in compound documents in addition to being separately downloadable, and should be able to be implemented as a component of the user-agent rather than of the underlying operating system. Many (most?) browser manufacturers have either announced or have already shipped initial support for native SVG. However, very few SVG implementations include support for SVG fonts. This may well be a chicken-and-egg problem, as I believe one of the primary use case for SVG fonts would be webfonts. Making SVG fonts the only _required_ format for compliance would avoid issues associated with favoring a particular vendor's patented technology, may reduce concerns related to DRM (since there are not thousands of proprietary SVG fonts out there to steal), and would limit referenced standards to those that are open and well-defined. Joshua RANDALL <joshua.randall@francetelecom.com> Senior Research Specialist France Telecom R&D, Boston
Received on Wednesday, 26 April 2006 00:11:06 UTC