PF/HTML_Caucus telecon -- 16 Oct 2009

<scribe> agenda: this

<Stevef> Recommendations regarding Long Text Alternatives http://www.w3.org/2009/06/Text-Alternatives-in-HTML5

<cyns> SF: alt text. I looked back at the wai concensus doc, and it reflected what I was saying in the WAI PF meeting yesterday

<cyns> RS: I hate longdesc

<cyns> JS: I think we agreed to obsolete longdesc

<cyns> SF: that path requires behavior for aria-described-by that weren't agreed to by ppl like Rich

<cyns> RS: in the accessibiltiy API, all described by is is a relationship that will take you somewhere in your doc, anywhere you want it to. I hate the idea of putting a link in there and forcing the browser to take the link.

<cyns> JS: force, or offer to the browser?

<cyns> SF: I thought it was just an offer

<cyns> RS: ok, that's better.

<cyns> SF: in the case where the longdesc has ref and id that's on a link, and the link text is further description, provide a mechnaism for the user to be albe to navigate teh link. A mechnanism whereby te user can move to the structured content

<cyns> RS: no problem with taht

<cyns> SF: a number of ID references could be a problem. Describedby is just supposed to concatenate text from multiple references.

<cyns> RS: if the content is hidden, tehre's nowhere to go to. it'll be in the dom but not in teh API

<cyns> RS: if it's hidden, UA will fill in MSAA description with whatever text it can gleen. It's like help text, can work with tooltip or whatever. But, the current interface in JAWS provides a mechanism to go to the descriptin.

<cyns> SF: The way to go there needs to be specked out for the AT vendor.

<cyns> JS: because longdesc was not a success

<cyns> RS: what about if there's more than one IDREF for description? That's not spec'd, and needs to be.

<cyns> SF: we could provide a set of steps. EG if the ID is on a table, move to table, if it's on a link provide an option to activate the link, if it's just a text container, just read out the content.

<cyns> MC: just allow normal processing, just go to the described by section. if there's a link, the user can click/activate it.

<cyns> RS: we can't make AT vendors do it.

<cyns> CS; no w3c specs for AT, hole

<cyns> RS: there is history, reasons for this.

<cyns> RS: AT vendors wouldn't follow first UAAG spec

<cyns> SF; need to revisist what we've said in text alternative document

<cyns> SF: provide idea in implementiaon guide on how we think AT should implement it.

<cyns> CS: put AT guidelines in UAG? or separate doc? good idea, but increase in scope

<cyns> RS: good luck getting them to do it.

<cyns> CS: took 10 years with browsers, maybe time to start with AT?

<cyns> JS; existing mechanism (longdesc) is a failure, look at why, build from that

<cyns> SF: some AT and browser vendors did implement longdesc

<cyns> SF: if we provide a better mechanism, they may implement it.

<cyns> JS: part of why we like ability to point via href via aria-describedby, is that we could do various media, not just text

<cyns> SF: if you have a long text alternative relationship, if there are 1 or 10 things you're pointing to, the user can get access to those things

<cyns> JS: jst like any 1 or 10 hrefs on a web page, no a special case

<cyns> RS: can you give me an example where you'd need more than one described-by?

<cyns> SF: one image that's more than one table? A graph, a data table in html that provides the data, that's the sort of relationship that a user would want to know about

<cyns> CS: multiple alternative versions for cognitive, like a picture, a video, a shorter version of the text, etc.

<cyns> RS: so, we need to say that if there are multiple IDREFs, each is an alternative description. AT can then bring up a list box, and you can pick the one you want.

<cyns> SF: agree

<cyns> JS: list should tell you want mime type it is

<cyns> RS: would have to be in a target

<cyns> CS: sounds like a may

<cyns> JS: I'd accept a MAY, or a SHOULD

<cyns> RS; do we want to put in in the spec?

<cyns> what about mainstream UA? should we ask them to do that?

<cyns> CS: yes, I think so

<cyns> RS: in the spec, make that a should or a may for now. When IE or FF does it, HTML5 will follow

<cyns> resolved: Multiple IDREF in an aria-describedby MUST be interpreted as multiple alternative rendereings, AT and other UA may expose mime-type to the user, non-AT UA SHOULD provide a mechanism for navigating to the IDREF

<cyns> RESOLVED: Multiple IDREF in an aria-describedby MUST be interpreted as multiple alternative rendereings, AT and other UA may expose mime-type to the user, non-AT UA SHOULD provide a mechanism for navigating to the IDREF

<scribe> scribe: janina

js: open agenda today, any topics for us?

rs: someone should be designated to start submitting changes from our spread sheet

cs: the mappings?

rs: yes, we're taking ian's and redoing it, so create an issue, make this change, etc

cs: i was thinking putting the entire table in as a change request

rs: were you planning to cover this on aria call?

cs: could

rs: should probably run on the aria group

cs: didn't we decide this at our f2f?

js: for our santa clara f2f?

cs: would like to get it to html wg before then
... at the very least i'd like to have this ready to submit by f2f

rs: i'll try to add some of the external dependencies

cs: frank fro ie team working on a canvas proposal, related, needs rich's guidance

rs: is there likely to be an html6 on the heels of html5?

cs: certainly a discussion

rs: instrumenting an a11y api for html5 will be almost impossible -- re canvas
... we can add dom, but that's aria2 --

cs: canvas is so different from html

- DRAFT -

PF/HTML_Caucus telecon

16 Oct 2009

Attendees

Contents

Summary of Action Items

Scribe.perl diagnostic output