The meeting produced a number of lists. The list of issues where agreement was reached:
David Meltzer, Microsoft davidm@microsoft.com Phil Karlton, Netscape, karlton@netscape.com Ashok Saxena, Adobe, asaxena@adobe.com Håkon Lie, W3C, howcome@w3.org Alex Beamon, Apple, alexbe@applelink.apple.com Jeet Kaul, Apple, jeet@apple.com Chuck Rowe, Agfa, chuck_rowe@agfa-type.com Robert Stevahn, HP, rstevahn@boi.hp.com Chris Lilley, W3C, chris@w3.org Randy Polen, Adobe, polen@adobe.com Steve Zilles, Adobe, szilles@adobe.com Glen Rippel, Bitstream, glen_rippel@bitstream.com Dave Raggett, W3C/HP, dsr@w3.org Erik van der Poel, Netscape, erik@netscape.com Ben Bauermeister, HP, ben@sea.hp.com Terence Dowling, Adobe, dowling@adobe.com
Discussions are mostly in chronological order
Chris welcomed everyone and -- after a round of introductions -- presented an initial outline of the agenda for the day. Also he presented an initial draft of a charter for the working group:
"The goal of the W3C working group on fonts is to produce recommendations for open font solutions for the web.Issues to be resolved include: style sheet notation, client-server request structure, and font location."
It was noted that the list of issues was not complete and that one should discuss the issue of style sheets vs. other notations before accepting it as an item of the working group.
DR: We need a request structure and it should be decoupled from HTTP. What information do one need to specify? The request structure should be format independant, but what comes back may not be.
The meeting started adding potential work items to the list:
SZ: Some of these issues are addressed in the IETF i18n (internationalization) draft, has everyone seen it?
The time scale of the work of the group was discussed. GR: we have to move fast to avoid fragmentation. We should focus on a few points.
A "fall rollout" was proposed as an initial goal with deployment (i.e. implemented by major browsers) by year's end.
The meeting started sketching a list of high priority issues:
PK: people are tired of sending large images across the web. We should show them how fonts on the web could replace many images. We need to come up with good examples.
RP: we need to understand customer requirements (content providers, content recipients) -- how much control do we want, how does the rest of the web fit into this? SZ: can we use the Microsoft/Adobe "WebFont" draft as a basis for further discussions?
There were no objections to this proposal. Since WebFont is partially based on CSS, HL started with a short presentation on CSS, including the @-notation for future extensions.
PK registered two concerns:
SZ presented WebFont which introduces the @fontdef notation in CSS to "define" (i.e. summarize) the font as well as pointing to a font resource on the web. It also proposes a new HTML element (DEF) to contain font definitions anywhere in the document.
The proposed DEF element was discussed. It was argued that one needs a way to do do font definitions on the fly inside HTML documents.
HL: @fontdefs could be put inside STYLE attributes but with global scope.
PK: this would be confusing as the normal CSS rules have scope.
It was suggested that the STYLE element should be allowed within the BODY element.
HL: this would lead authors to think one could include CSS rules with gobal scope inside within the document.
PK: perhaps this is a good idea, perhaps with a downward scope?
A discussion on the copyrightability of font metrics emerged; the answer is probably yes.
BB presented panose 1 & 2. HP is looking for a new home for Panose. Panose 2 was favorably received at the meeting, incorporating it into OpenType would require changes to the specification.
Can Panose live next to PDF-inspired fontdefs?
BB: not next to, perhaps one could combine. Panose numbers have also been used to synthesize fonts.
The discussion turned to glyph widths which are optional in the WebFont font descriprion. There are two reasons for having them: progressive rendering and a basis for making a signature.
It was discussed if font definitions could/should be used in the request structure.
<lunch></lunch>
CL: what have we achieved up to now? What are our goals?
A list of goals was worked out:
GR: we should make a statement on the font face issue, should we do style sheets, or HTML elements/attributes?
non-goal (in phase 1 & beyond): not solve all IP issues,
Should W3C accept font donations? There was various feedback on this issue, most people seemed to think a document with pointers to font resources was sufficient.
SZ: what are the difficult issues?:
It was generally agreed that HTML was not the right tool for ensuring absolute layout, HTML is not PDF
DR preseted ideas on fonts and caching. Is it worth doing subsetting for non-cjk fonts? The answer is not clear.
SZ summarised the communication that would go across the wire.
PK: I assumed one could include font outlines in the font definition, no?
The current WebFont proposal does not allow for this. It was argued that the "data:..." notation could be used to include font outlines inside HTML.
The discussion turned to deliverables: what are the documents we should produce in the working group? The meeting first split deliverable into 3 documents and then assigned people responsible for an initial draft:
The initial drafts should be ready by July 1st. They will be partly based on the "WebFont" document.
Font signatures were discussed. It was generally agreed that this was a useful way of identifying a font.
DM: PC and Mac binaries of the same font may be different: there may be information in there to identify source, and various compression techniques may be employed on different platforms.
TD: font face + glyph complement may be suffisciently unique
How are fonts installed on systems?
DM: Through an API, I understand if that scares some people
Should one specify the font face as an attribute on the FACE element on the FONT tag or through a new CSS property?
PK: CSS or not, we want a solution! If CSS is there and works, we're happy to use it.
GR: is the only way to get the font url through the font definition?
Yes. LINK will not do this.
The meeting reached agreement on the following issues:
w3c-fonts-wg@w3.org will be the mailing list of the working group, while www-font@w3.org is the public alternative. Phone meetings is another alternative: 1pm california = 10pm france.