WebFonts ad-hoc planning meeting
05 Nov 2009


See also: IRC log


Sylvain Galineau (Microsoft), Chris Lilley (W3C), Jun Fujisawa (Canon), Tab Atkins (W3C Invited Expert), Liam Quin (W3C), Vladimir Levantovsky (Monotype), Doug Schepers (W3C), Christopher Slye (Adobe), Arron Eich (Microsoft), Bert Bos (W3C), Arun Ranganathan (Mozilla - irc)
John Daggett (Mozilla), Steve Zilles (Adobe)
ChrisL, TabAtkins




<TabAtkins> ChrisL: Apologies for not attending in person. I'll be here in irc; i'm just over in html right now.

<ChrisL> Tab, no worries

<TabAtkins> You're not the chair, Sylvain. You don't get to assess penalties.

<TabAtkins> ^_^


<TabAtkins> I'm assuming there are more than you two in the physical room right now?

<sylvaing> Tab: yes. Vladimir Levantovksy (MonoType) and Christopher Slye (Adobe) are also here

<TabAtkins> Hmm. I might come over then. Just a sec.

Charter review

<ChrisL> http://www.w3.org/2009/08/WebFonts/charter.html

<ChrisL> SG: attribution of EOTLite - its an ascender proposal not John Daggett

<sylvaing> Ascender's official announcement: http://www.ascendercorp.com/pr/2009-07-15/

<ChrisL> SG: John and I fleshed out the details based on the ascender proposal

<ChrisL> SG: I offered to write it up for a submission, great to have John as co-author

<ChrisL> ... and available to edit

<ChrisL> SG: what does scare quote "DRM like" mean

<ChrisL> CL: reference was to previous, failed attempts

<ChrisL> SG: Its acknowledging FUD, so please remove

<ChrisL> CL: wanted to make totally new formats out of scope

<ChrisL> SG: Why single out EOTLite, SVG fonts and raw OT are also questioned

<ChrisL> scribe: ChrisL

<TabAtkins> slye: I don't see why EOT is more out-of-scope than any other format

<scribe> scribe: TabAtkins

vlad: Get browser vendors, users, etc comfortable is the goal.

ChrisL: It's obvious woff is the one we want, and so people in the group have said "just do that".

sylvain: That's been mostly mozilla pushing back on eot/for woff.
... not so long ago we ehad people saying the only thing that is needed is raw fonts
... so we want feedback on all of them

<ChrisL> http://www.w3.org/2009/08/WebFonts/charter.html

chrisL: Anyone have problems with the charter that aren't already planned on being discussed in the agenda?
... Tab, to be clear, it's not *me* who's saying "WOFF is the one", it's just what people have said.
... Next topic in agenda - which formats to standardize?

Which formats to standardize

TabAtkins: First obvious question: Should we just do all of them?

Sylvain: We could, but we have to define all the conformance criteria.
... There has been chatter between me and ??? about issues not directly tied to the format, like access policies.

ChrisL: So you'd like to see access policy become an explicit deliverable?

sylvain: Yes. Right now 3 browsers implement raw fonts but have different access policies, which is confusing for authors.

ChrisL: On that note, encourage font vendors to have *clear* licenses that explain the font's use as a webfont.
... Right now contracts tend to be directed toward desktop publishing, and are unclear on what, frex, 'embedding' means.
... Even if someone has the budget and the will to use a font, it can be difficult to still figure it out without 1-to-1 chatter.
... If a bunch of foundries were even to say, "You can't use these fonts on the web", that would be great, because at least it would be clear in practical and legal terms.

Slye: Adobe's very much aware of this.
... It's complicated by our sublicensing with other foundries. We intend to get that clarity eventually.

ChrisL: Yeah, it's related to how you got the font. Did it come with a software, get given as a sample, etc.

Vladimir: There's the complicated licensing structure of fonts. There are 160k fonts sold by Monotype, but not all are owned by us, so we can't just unilaterally change licensing.

ChrisL: Possible deliverable - note to encourage clearer license, perhaps examples?

Vladimir: I believe strongly in market forces. It will be valuable when authors have a multitude of choices, so the license relationship affects how successful the foundry will be. I'm not sure the WG should try to regulate this.

Chrisl: I'm not thinking of anything normative, just an example, frex for foundries who are just now coming into the webfont space.

Slye: One problem with that is that I see differences of opinion about best practics. A lot of people ahve different attitudes about it, authors vs corporate lawyers frex.
... You look at something simple that everyone likes, then the lawyers come in and carve it all up.
... It sounds good to promote best practices, but I'm not sure it'll work out in reality.

ChrisL: Absolutely, not a conformance requirement. Just "consider this, consider that" to help point out areas that are good and bad.

Vladimir: I can say on bhalf of monotype that we take this seriously, and if there is a single author that walks away fontless because of the unclarity of the license...

Sylvain: On the page with the buy button, do I get a clear idea of where I can use the font?
... No, it's buried down in the package.
... If you want market forces, you need a market. With fonts right now, it's not clear what you're getting.

Vladimir: I can bring up the same complaint about every piece of software on the web.

Sylvain: That doesn't mean we need to stick with that just because it's how it's been done.

TabAtkins: stock photo sites can do this clearly, even if there is legalese sitting behind it.

Slye: [something about embedding bits, fonts in pdf]

ChrisL: I think altering the OpenType spec is beyond what we want.

Slye: Yeah, but just keep in mind what might be useful.

Vladimir: SC29 can establish liaison relationships. I can establish that.

ChrisL: We already have a class C liaison from sc29

Sylvain: The last thing about licensing is making it clear for authors.

<ChrisL> Vlad is the editor of the relevant part of MPEG which standardises Open Type (OFF) and can be the liaison to them

Sylvain: It is a major problem that authors don't understand the issues with licensing. They think "It's in my font folder, I can use it on my myspace". This is a problem, and it needs to be clear.

TabAtkins: I'd appreciate calling attention to it in the spec, even if there's no suggestions, so that everyone is aware.

Sylvain: Even free fonts, I'm not sure if I can put them on a webpage. I still need to check the license.

ChrisL: Even libre fonts - the most common collection has a restriction that you can modify, but you must rename. You can't do *anything* with them.
... Even converting between formats (ttf to woff, frex) may break the license.

Sylvain: Being able to see what is and is not allowed in clear terms before downloading is a valuable thing.

Chrisl: Bringing the topic back to which formats to standardize.

Sylvain: We need to standardize and document the two that are not. The main argument has been about the conformance criteria.
... They think it doens't go far enough, even if they know in practice that it'll give a good result.

TabAtkins: If we know in practice that "2 of 4" will give us an interop format, let's just specify that format.

Sylvain: Tab, as an author, are you okay with having to make, say, woff in one browser, eot in another...

TabAtkins: No, it's just as crap as it is right now. I know this won't be fixed until later.

Sylvain: But it'll still be better, because frex firefix can only use raw fonts right now, and you can't get fonts out of, say, monotype that are allowed.

ChrisL: So what I'm hearing is that the charter should mention both formats as work items, which it does currently.

Sylvain: If we pick one format, the benefit of standardizing the other is mostly theoretical.

Vladimir: I think it should be up to the working group, not the charter, to decide that.
... After good progress in discussion over the summer, I'm a lot more optimistic about the outcome.

Sylvain: Frex, say we have 25 font formats on the table, 20 undocumented, and the conformance is going to be 1. Do we really *want* to document the others?

Vladimir: I'm specifically supporting two.

Chrisl: Both new formats are just transformations of existing formats, so the spec will be relatively short anyway.

Vladimir: CWT is literally 1 page.

Sylvain: CWT as specced is not implemented. It's compatible with IE.
... I don't want to write a spec if the conformance criteria says that the format won't get implemented.
... Are we going to write specs first and then pick?

Vladimir: I dont' think we can establish a sequence today or in the charter.


Sylvain: Then let's talk conformance criteria. Assume the specs will be written.

ChrisL: The original fonts spec specified the font-linking mechanism, but said nothing about format, and so it failed.
... SVG did mandate one format, and got interoperable implementations.

TabAtkins: I don't think there is any technical consideration from waiting for complete specs that will sway the issue.

ChrisL: But the browser support issue may change in a few months.

TabAtkins: If we're wanting to wait for impl xp, I'd want to be precise about that. The charter should say "We're waiting, and at XXX date, based on browser xp, we will choose the specific formats that must be supported."

Sylvain: Not certain if the browsers will be happy about that. I'm the only browser rep here.

Vladimir: Interop in the future is the cake. Interop now is the icing on the cake.

ChrisL: You have to know you're going to get there. XHTML2 planned for the future, but didn't worry about what would happen if nobody came to the party.

Sylvain: If MS shipped WOFF without same-origin tomorrow, it would be toothless. Almost all EULAs require that sort of checking.
... Interop must be defined widely, not just bits on the disk.

Vladimir: Agreed, and I think the solution is the one that hits all the people here.

Sylvain: Saying "We pick XXX" is the most spineful solution, the ideal, but it may be ignoring the reality of today and blowing away a realistic good-enough solution.

Vladimir: I think the charter should say the WG should define conformance.

Sylvain: But a lot of people don't want that - they want the conformance criteria set down in the charter.

ChrisL: The charter can say what it wants, but if it says something other than what the WG wants to do, that's a problem. So we need to align it in the direction we want, leave enough wiggle room for the future, and define everything else out-of-scope to keep us on track.

Vladimir: My concern is that if we leave it as it is, 2 of 4, there will be little reason for vendors to come to the table. They might be conformant on paper (otf, svg), but nobody is helpeed.
... If the charter says the WG will decide, then they have to come to the table.

ChrisL: Can we just say in the charter that the conformance criteria will be decided by the WG?

TabAtkins: Some bits of conformance criteria (same-origin, frex) should be specifically noted. Otherwise yes.

Slye: So are we going to say "Here's a set of formats we're looking at, we will decide for certain which of these will be required in the future, but it'll definitely be one of these.

TabAtkins: Yeah, I'm okay with that.

ChrisL: In practice, we know for a fact that there are only these two formats. SVG is useful in its way but not a complete solution, and raw fonts just make too many people angry.
... So out of the 4 that have any implementations, WOFF and CWT are the true choices here.

[chatter about CWT naming]

ChrisL: So specifically, I think the "2 of 4" needs to go, and needs to say "one or both of WOFF and CWT".

TabAtkins, Vladimir: Yeah.

ChrisL: [talk about timeline]
... So publish a FPWD and say "Here are the choices, ideas?" and then publish another WD with the decision.

TabAtkins: Says yeah.

Sylvain: Instead, say "both" right now, and note that in the future we will, if necessary, just do one.

ChrisL: I don't want the charter to founder on someone saying "I'm doing EOT over my dead body".

Sylvain: I need other browser vendors in the WG.
... I don't want Daggett, Hakon just staying out and picking one format through inaction.

[talk about rootstring-in-padding, same as list]

Sylvain: I wanted the no-rootstring version of EOT, which helps the legal issues, but then you have licensing problems.

Vladimir: Mozilla was against EOT. They used DRM as an argument, they used patent..

Sylvain: Patent issue was real; it was a chicken-and-egg issue. They were fair arguments.

Vladimir: Apple just pushed out a big pack of patents for CSS specs, and Moz didn't say a word.

Sylvain: But you can implement transitions without infringing, but MTX has no alternative.
... There were definitely GPL/field-of-use issues.

ChrisL: Conformance criteria!

Sylvain: Looks like we don't want to say "do two" because it'll get objected at the AC, probably by Moz/Opera.
... ROC said specifically that the WG should only bless WOFF.
... I have the email. (If I said that there'd be someone on the phone to the EU in seconds.)


Sylvain: Simon Daniels has been gathering feedback on formats, and MS has considered raw fonts. We have stated publicly now that we would do it only with same-origin, and the font must be installable (no embedding bits are set at all).

ChrisL: Silverlight has a different policy?

Sylvain: SL has changed. It doesn't do docs, it does apps, which is a different legal issue?
... Essentially you couldn't grab a font from your font folder and upload it, because the embedding bits would be set.
... If we were to do raw fonts, that would be the condition.

ChrisL: Would you do WOFF under certain conditions?

Sylvain: We're looking at it, but we want to see more than waving banners. We want to see licenses.
... A license that specifically sells you WOFF, or lets you encode it as WOFF.

Vladimir: I'm pretty sure my license will be modified to allow WOFF
... It will take time to get through legal, of course. I'm pretty sure other font vendors will do the same.

<arun> yay Vladimir :)

Vladimir: This is the thing we want market forces to shape.
... Lawyers want strict definitions. If they'll use WOFF as a term, it *must* be defined somehow.
... There are certain steps that have to be taken.

Sylvain: For similar reasons we'd like it as a w3c spec.

Slye: Has Moz said anything about MS's talk of raw fonts?

Sylvain: We've talked to everyone, and there hasn't been ngative feedback.
... You can only really use free fonts, unless you hack it, which we can't prevent.

Slye: But the way you describe rawfont today, they're not interop.

Sylvain: You can use free fonts in Moz and future IE in the same way.
... We probably want to define rawfont conformance/interop.

Vladimir: In one way, we can say that embedding bits that exist today are not applicable to web linking. ??? spec would need to be changed.

Sylvain: We're saying that *no* embedding bits means you can use it (no restrictions).

TabAtkins: I'd like to change my answer. I think that rawfont embedding is philosophically tenable for the libre community, and practically tenable for foundries.
... You'd have the 'garden fence'.

Sylvain: *All* commercial fonts have the embedding bits set.

<br class=for-real>

<ChrisL> [break, 15 mins]


ChrisL: [recap of w3c process, testing]
... It's good to have the testing directly in the charter.
... We can talk about generator/consumer/browser etc conformance.
... There is precedent for produceers to push out oddly incompatible formats, and consumers to only accept output from certain tools, etc., so testing along the chain is a real concern for fonts.

Sylvain: So when do you make the format decision, because that will affect the test suite.

ChrisL: We've had cases where downloadable fonts are 'supported', but you have to do weirdness with bold/italic/etc as different faces.

Sylvain: For the purposes of this WG, how do you decide browser X supports font A?

Vladimir: You put the font in a @font-face, and check what renders.

Arron: And check if it downloads (http traffic), and then do a visual check to see if the actual font is being used.

ChrisL: SVG has similar tests with visual checks. It's obvious when you pass/fail.
... You mentioned that IE8 doesn't support CSS3 Fonts?

Sylvain: We do a good subset of CSS2 Fonts, but not CSS3, like unicode ranges.
... And the optional format string, we don't do that.

ChrisL: But IE8's already shipped. What's IE9's expected shipping date? 2 years?

Sylvain: Yeah.

<sylvaing> correction: our typical release cycle is 2 years.

Arron: shipped in March '09

Bert: So we won't see it early next year.
... Some WGs split up into different conformance groups.

Sylvain: That's an argument against EOT - even if other browsers implement it, IE's lack of full CSS3 Fonts support may still cause problems.
... And that might come up in AC.

TabAtkins: There've been experiments on how to work around IE's limitations while still serving to other browsers.

[Liam Quin enters the room.]

Bert: What's XSL-FO's relationship to webfonts?

Liam: We put out a FPWD in Sep, and basically say we're waiting for the webfont discussion.

Bert: Do you think we'll have impl?

Liam: Yes, we expect so. [Something about FOP? XMLROF?]

ChrisL: XSL-FO and SVG both would benefit/are waiting for webfonts.
... To have success, there must be ways to produce fonts, there must be licenses and fonts for sell, there must be browsers that consume properly.
... Format tests - testing if a tool can produce a font in the right format - will be some of the easiest parts.

Slye: If WOFF becomes so successful that it's a virtual desktop font - the point of WOFF was that it was *not* a desktop font.

Sylvain: Like if MS says "WOFF is supported by the OS now".

Slye: We don't want people downloading fonts from the web.

Liam: It'll definitely happen in Linux.

Sylvain: It's technically logical to build format support directly into the OS, so different programs don't have to reipmlement for themselves.

Liam: The hope is that it'll be at least a year or two before OSes start doing this, so we can get some education.

Vladimir: WOFF also has specific places for user-specific info.

Sylvain: But no physical restrictions based on that.

Vladimir: Yeah.

Liam: Have we talked about UI? We'd want, frex, browsers to possibly have a view of what fonts are used and what their license is. We don't want to require UI, but perhaps in a plugin or something.

Arron: LIke right-click on a font and select "Font Properties..." or similar.

ChrisL: As Liam says, it's a marketing opportunity. The web means better exposure for fonts and type designers.

<ChrisL> whether its libre fonts or commercial fonts

Liam: Part of the goal of WOFF is just to get it *used* on the web.

Sylvain: I can see web designers seeing a cool font and pulling up Firebug to see what it is and where to get it.

ChrisL: Anything more to say about testing?

TabAtkins: If we say specifically that we'll do testing on both browsers and generator tools, I'm okay.

Slye: Digital signatures in fonts - have they been discussed?

ChrisL: Not yet.

Vladimir: DS is a *component* of a font that will be dropped into WOFF or CWT. So I think that it belongs to a different thing.

ChrisL: There was talk that WOFF should be decodable back to a rawfont that matches DS.

Vladimir: Specifically, say that the font should come back out of the format identical - don't reference DS specifically, because it's optional.

Sylvain: Are commercial fonts commonly signed today?

Slye: Yes, for Adobe.

Vladimir: It's per-vendor, basically.

RESOLUTION: Test that it's possible to decode a font from CWT and WOFF back to an identical copy of what was encoded.

Rather, ensure that this is part of the testing we do.

Fonts MIME type

ChrisL: Just to make people aware that there is a font mime type about to be registered.

Vladimir: The document is still waiting to get more support behind it, so IANA has a harder time shooting it down.

ChrisL: I think w3c should be behind this effort, and there has before been a top-level type defined, "Model" - for 3d/vml.
... It's happened in the past, so it may happen for us.

Vladimir: What's changed is that fonts have become reasonably common components that travel along with the process and present security contexts.
... Before there wasn't a lot of *reason* for it, because fonts didn't move betweeen domains etc.

ChrisL: It's good to note the security issues directly, because bad fonts can crash your OS. It's important to note that in the proposal.

Vladimir: Yeah, it's fairly easy to hide things in the font that doesn't belong there. It's almost too scary.

Sylvain: Yeah, hinting is basically a program.

Liam: I could see an argument for application/font, because it runs code.

Vladimir: No, it's less of a problem because it's code that's interpreted in a sandbox.

Liam: That worked for javascript!

Sylvain: I think adding random data to a font won't *do* anything directly, if nothing is going to read it.

Vladimir: Still, it's information you don't know about stored on your system. It's a possible attack vector.
... So top-level font mime type is a good thing?

Everyone: Yeah.

Slye: What's the *advantage* of a top-level versus not-top-level?

ChrisL: It's more obvious what you're getting.
... People don't like images/etc all coming down as "application/*".

Liam: We have a w3c process where a spec can have a mime-type registration. It still has to go through ietf, though.

Vladimir: [discusses current mime types for fonts]
... I'm editing part 1 and 2 of sc29/wg11

ChrisL: I want a strong liaison, because that's a way of heading off the group from doing crazy astro-engineering.

First ftf meeting

ChrisL: It's possible in the charter to say "If this is accepted, we'll meet xxx at yyy."
... But there are travel restrictions that say "one ftf per year".
... So we can do it early or in several months. thoughts?

Slye: is the ftf closed or open?

ChrisL: Closed to the WG only, but worked in public and minutes are published.
... I'm hopeing we can be reasonably liberal about invited experts.

Vladimir: Like Ascender isn't a member company.

Liam: That'll be difficult.

ChrisL: Big organizations should really join the w3c. Invited Experts are more for small shops that have usful expertise, but not the budget for full membership.
... So thoughts about how early to have the meeting?

Vladimir: What's the earliest reasonable time-frame, given the charter will take a while?

ChrisL: I expect the WG to start beginning of Jan.
... So do we want a kick-off meeting?

Sylvain: Do we want to colocate with CSSWG?

Bert, ChrisL: Sure.

Vladimir: When is that?

ChrisL: Week of the 29th, 29-31.


[Doug Schepers enters the room.]

XSL-FO also has a meeting in March, in Prague.

Doug: [talk about wrapping text to shapes, Liam will talk with Doug about XSL-FO's practical experience in the matter]

ChrisL: So, Cupertino. Do we have a host?

Slye: Adobe may be able to.

ChrisL: I'll talk to Steve about it.

<jun> On top-level font MIME type: http://lists.w3.org/Archives/Public/www-style/2008May/0254.html

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2009/11/06 22:58:25 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.135  of Date: 2009/03/02 03:52:20  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: s/level 3/class C/
Succeeded: s/political/theoretical/
Succeeded: s/got in trouble with the W3C/got interoperable implementations/
Succeeded: s/ChrisL/Vladimir/
Found Scribe: ChrisL
Inferring ScribeNick: ChrisL
Found Scribe: TabAtkins
Inferring ScribeNick: TabAtkins
Scribes: ChrisL, TabAtkins
ScribeNicks: ChrisL, TabAtkins
Present: Sylvain Bert ChrisL Jun Tab Liam Vladimir Doug Slye Arron
Agenda: http://lists.w3.org/Archives/Public/www-font/2009OctDec/0173.html
Got date from IRC log name: 05 Nov 2009
Guessing minutes URL: http://www.w3.org/2009/11/05-webfonts-minutes.html
People with action items: 

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.

[End of scribe.perl diagnostic output]