WebFonts Working Group Teleconference -- 22 Jun 2011

<trackbot> Date: 22 June 2011

<scribe> scribe: ChrisL

<erik> can't call in, sorry

Formal objection from Samsung

http://lists.w3.org/Archives/Public/www-font/2011AprJun/0069.html

Vlad: Glenn has joined the call.Please give the background for your objection. There has been much mailing list discussion

glenn: represent Samsung.Padt member of CSS WG.
... in section 1 intro it has three requirements on user agents
... unusual to have normative requirements in introduction
... the note seems to contain a normative must, unusual in a note
... these should not be in an introduction
... core issue is that these three paragraphs and note make reference to css3 rules and ua behaviour,this constrains implementations of woff
... other implementations might use other ferencencing mechanisms or other access policy
... majorr objection is use of referencing mechanisms and ua resource fetching in this file format specification rather than some other document
... if these are all removed and subtiture text offered this would solve the objection

jdaggett: which version are you looking at

<Vlad> http://dev.w3.org/webfonts/WOFF/spec/

<cslye> http://dev.w3.org/webfonts/WOFF/spec/

ChrisL: the /TR version. Glenn, the editors draft has moved this normative text from the introduction

jdaggett: which version of the css3 fonts spec did you look at?

glenn: just the woff part
... ok so my first issue is solved
... in the editors draft
... want to see the general requirements removed
... a separate document defining this would be okay
... access control policy that applies to a ua or other agent that uses this file format

jdaggett: on the list you seem to want the css3 font spec to change also? or only the woff spec

glenn: same comments on css3 font spec it should not define the fetching process
... the way to refer to a font. not a fetching mechanism. nothing says about the transport protocol used to fetch this
... this wg does not develop css3 fonts

Vlad: yes but they are markjed as features at risk because we believe they should be removed from the format spec and placed in css3 fonts instead

glenn: understand but dont't agree
... right nw we have looping references
... want a tree graph not a circular graph

ChrisL: your mail says that moving all this to html5 how would that solve your problem?

glenn: no preference

ChrisL: html5 spec and css3 spec both define auto fetching linked resources, how do they differ

glenn: html5 defines the fetching mechanism in detail
... css3 fonts does not define that mechanism not should it

jdaggett: so you are sayingit cant define requirements around that fetching?

jfkthame: throughout the woff document there are ua requirements

glenn: png or mpeg or jpeg dont define ua requirements in the file format. format should be independent. conformance on processing is reasonable, like encoding or decoding
... conformance levels related to presentation.
... those are reasonable
... requirements on access mechanisms and transport protocols are not appropriate

jfkthame: we agree the file format is not the place to define it. that was accepted and agreed but it is not as yet defined anywhere else
... much harder to understand your objection to defining that in css3 fonts. its defining the @font-face rule and entirely appropriate to define how a ua behaves when processing that rule

glenn: other referencing specs at w3c like xlink or xml stylesheet spec or css 2.1 which refer to other resources dont define fetching or access semantics
... so that precedent should apply here

jdaggett: seems to be a thoretical distinction. why the requirement to follow those boundaries

glenn: group can do as they wish, define different levels. specs are mixing layers here, transport and formats

jdaggett: do you feel the same way about the loading mechanism of images in the canvas api, like the tainting rules - do you object there? this is very similar

glenn: canvas started in html5
... its linked to html5 in a way that css3 fonts is not. it doid not start out by having acccess mechanisms in it

Vlad: historically css started as part of the html activity then migrated

jdaggett: canvas has a similar definition about origin, if the canvas references images from a particukar origin it has impact on ua behavioour. its part of the html5 spec right now at w3c.
... the reason its erelevant to this discussionis an example is it defines ua behaviour

glenn: html5 is a definition of UA behaviour. css 2 does not define origin requirements

jdaggett: are you saying no definition of access can be in css specs?

glenn: html5 defines a user agent. css3 should be referenceable by other specs that use other acess control or transport mechanisms

jdaggett: cant see how a spec that defines fetching resources is not a user agent specification

glenn: css can and has been employed in other contexts. idea of modularisation is to make specs independent. fetchingsemantics in css3 is going backwards. unnecessary dependencies that are not rwquired in woff or css3

Vlad: are you saying that if we had this text in a separate spec is okay

glenn: yes

jdaggett: pushing everything out to another spec makes no sense its findamental to the @font-face rule

cslye: it was decided after long discussion that it was relevant

glenn: html 5 has a section that goes into great detail on resource fetching. that is a good place to define this also
... or in a separate spec
... work with authors and content providers (scribe missed some) guidance to authors and we are folliwing up on this. if these requirements remain then our specs will override this and make them optional in our specs
... if this is in a separate spec we might reference that in some profiles

cslye: so this seems to undermine the point that defining this is generally inappropriate

glenn: after looking at the email again we dont object to same origin per se or to SOR vs From-origin. want the option for another group that i am working with to have the option to refer to woff and to css3 font face and have the option to include sor or not

jdaggett: a group that is not epub?

Vlad: epub does not have confidentially restrictions

glenn: its a group for consumer electronics and the fcc has adopted its specs for tvs and handhed devices
... due to confidentiallity i can't say more

jdaggett: so this impacts creating a profile?

gl: yes, it means we have to overide it instead of the flexibility of making it optional

jfkthame; consensus of the group was that this should not be optional

cslye: yes that was what made the font vendors comfortable with it. adobe would see no value in this spec without sor

glenn: want this to be optional. we can overide it if you publish like this but we think its architecturally unsound. understand that the group has asked font vendors.
... neither truetype or opentype or pdf define this

cslye: woff is not a font format its a delivery container

glenn: yes

jdaggett: having specs refer to other optional specs mean aiuthors cant rely on the feature. so it makes things not work

glenn: might allow the user to disable user agent restrictions,that is another option

jd; we are primarily interested in specs defined by w3c. if other people want to profile this in other ways ...

glenn: dont see this is only for use for w3c defined user agents

Vlad: actually the group charter says that
... first statement, mission is for interiperable download of fonts on the web

glenn: ah okay

Vlad: you said that part of the group has a strong interest ... actually that is a resolution of the whole wg. and normative behavious gives interop, this is also a group consensus
... agree that the format spec is not iseal, best place is in css3 fonts which is where this referencing mechanism is defined

glenn; we will object if its in either of woff or css3 specs

Vlad: so primar y requirement is to be able to profile it out?

glenn: no the primary objection is the mixing of layers

cslye: do others support you on that?

glenn: yes i have had some supporting email. have not looked at other participants in this other consumer electronics forum but some of them are w3c members so i will ask what their position is
... if samsung is a lone dissenter we might drop the objection later
... will look at how this is resolved

jdaggett; as editor of css3 fonts spec, i dont see a way of changing the spec to what you are asking for, so that the same origin mechanism goes into a third spec. it merely pushes the specs around rather than adressing the actual issue of what the origin mechanism should be

glenn: understand your position
... see that html5 defines same origin and something related to fonts
... this is where it should be defined

Vlad: thanks glenn for joining us so we can better undertand each other's positions. this has beena positive discussion and i think we all understand the issues now
... can see that other organisations develop subset specs, this has happened before. that is fine but for w3c we want somthing that is coherent, tstraightforward as possible

glenn: on a final not, we are not trying to make a change thatprecludes content authors restricting access to content. no objecting to that. dont want to stop content authors or font foundries protecting their content or intellectiual property
... there are mechanisms for controlling access
... autgors can express constraints on access and uas can accept those contraints. no problem with that
... happy to attend a future call, thanks for the discussions

agdourned

- DRAFT -

WebFonts Working Group Teleconference

22 Jun 2011

Attendees

Contents

Formal objection from Samsung

Summary of Action Items

Scribe.perl diagnostic output