Protocols and Formats Working Group Teleconference -- 14 May 2015

<trackbot> Date: 14 May 2015

<scribe> scribe: joanie

<richardschwerdtfeger> https://lists.w3.org/Archives/Public/public-pfwg/2015May/0078.html

Heads Up Items

RS: Janina, when are you reading to talk about aria-describedat?

Janina: When you come back from vacation.

RS: I'll be on vacation for two weeks and will need a chair.

Heartbeat

RS: Issues?

MC: Not for this group.
... I did a bunch of cleanup to the documents.
... If I didn't think they were editorial, I pinged the editors.
... Hopefully they'll be up later today.

RS: aria-interactive didn't make it right?

MC: Correct.

aria-interactive

<clown> s/ARIA Interactive/aria-interactive/g

<clown> g/ARIA Interactive/aria-interactive/

RS: Matt, anything you want to talk about?

MK: I guess we didn't make the cut, but that's ok.
... It turned out there was a lot more to think about and talk about than I had anticipated.
... Do people have the link to the branch?

RS: No

<clown> http://rawgit.com/w3c/aria/matt-action1505/aria/aria.html#aria-interactive

Link to branch is above

MK: The first issue to deal with had to do with the desirable behavior of descendant children.
... As I understand it, aria-interactive:false on a grid would then make gridcells non-interactive.
... That would then mean we would not need a new cell role.
... Authors could just mark the property on the grid and the property would be inherited by the gridcells.
... And then the gridcells would like like a plain td to ATs.
... But you don't want all descendants to get this property; just the required owned elements.

RS: This applies also to rows, column headers, row headers.

MK: Are column headers and row headers required?
... They have required context.
... Is this going to be a nightmare?

JS: It should be the same as the presentation or none role.

CS: Can the author override that?

MK: No.

JG: If you're not exposing the widget role, how do you know what the appropriate non-widget role is?

RS: We'd do that through the mapping guide.
... Example, IA2 has a role of cell, but then they look at the object attributes for the grid cell.

CS: It would be mapped to however it's mapped on the host operating system.

MK: So it comes across as gridcell with aria-interactive:false.
... And that would map the same regardless of the host language.

CS: Yes, on a given platform.

JG: So if I put <div role="gridcell"> that will look like a td when aria-interactive is false?

MK: Yes.
... At the moment, we're only put this on grid and list.
... I changed the ontology.
... Previously, list had as subclasses menu and directory.
... And directory had tablist as one of its subclasses.
... I removed menu from list and tablist from directory. Now both tablist and menu are composites and do not inherit from list.
... Now aria-interactive will operate on grid, treegrid, list, and directory.

RS: Were there any inherited states and properties lost as a result?

MK: I don't think so.

JD: I had checked. The answer is no.

RS: Just double-check it.

MK: There's one more change if the group still wants to go in this direction.
... I can see both plusses and minuses to this approach.
... I'm not sure this will be better than adding new roles.

RS: I think people are already doing this anyways.

<ShaneM> Editorial comment: aria-interactive in "list" has an   and whitespace before it in the table entry

RS: For instance using grid as a non-interactive table. I've heard this is already happening in the wild.

MK: I think when you're nesting tables in tables, you really need to be explicit.... I cannot think of any situation when you'd want to have an interactive child inside a non-interactive parent or vice versa.
... For example, in a static table you don't need to make the sortable column headers interactive. Just make them standarrd links, right?

RS: It's less likely if you're using divs and spans.
... If someone wants to embedded a grid in a cell of a non-interactive grid, you are not prohibiting that.

MK: Correct.

RS: Did you look at treegrid?

MK: Not yet.

JG: Is aria-interactive considered a repair technique?

RS: No, I don't think so.

JG: But you're using grids, and if you put this on it, you're making it non-interactive.
... What's the difference between a non-interactive grid and a table?

MK: Nothing.
... This gives us a way to support tables in host languages which lack them.
... On the flip side, on a static element like list, this will result in a new interactive widget: a listview.
... That's a capability we didn't have before.

JG: So a gridcell can be interactive or not based on this property?

MK: A grid can be.

JG: This will be interesting to explain to authors.

MK: I was thinking about this. And this is one reason I'm not convinced this is better than new roles with respect to authoring.

RS: I think we're ok for now. Let's see what happens in the wild if people complain.

JG: Besides list and grid, how many other candidates are there?

MK: In my estimation, not many.
... Related, why isn't toolbar a composite?

JS: I agree with you, but this discussion was raised at least twice before.
... The conclusion was that a toolbar should be a structure.

RS: If you want to raise an issue, raise an issue.

<Zakim> clown, you wanted to note that practically any element can be made interactive via script.

<richardschwerdtfeger> https://lists.w3.org/Archives/Public/public-pfwg/2015May/0078.html

ARIA Extension Review

<richardschwerdtfeger> https://www.w3.org/WAI/PF/wiki/ARIAExtensions

RS: Shane and I worked on this a bit.
... Are people happy with this?

Janina: In the global thing, I think we maybe want to decide between "extension" or "module".
... The why is because what happens once it is accepted.
... On the user agent side, you have to implement it because it's part of the ARIA specification. Right?

CS: I'm not sure.
... Thinking out loud, there are pieces. For instance, patterns I think can be pretty separable.
... You could support one without the other and it would still work. You'd have holes, but it should work.

Janina: We need disambiguation somewhere (e.g. the "part" role).

Shane: I think anything that becomes a W3C REC (see item 4), then it's required.
... (Reads from his text)
... It should say "every conforming ARIA implementation".
... But that's my opinion, and not what Cynthia is saying.

<Zakim> ShaneM, you wanted to answer janina

CS: I'm thinking about how this will be implemented in software.
... It's a collection of features that fits into a milestone.
... If it maps to something, that makes it easier to implement.
... If it's a big monolithic spec, that's harder to accomplish.

RS: I want to separate spec from extension.
... If something can be implemented broadly, we can put that in a spec.
... If it's specific to a particular area/industry, like digital publishing, we'd put it into a module.
... It's also possible that the digital publishing would not go into a regular browser implementation; just a digital publishing version of that browser.

Tzviya: I think it would be helpful in writing the module, if we could leave the concept of roles being brought into the master specification aside.
... I don't want to have to keep track of which role is implemented where.
... I think it's great that roles might be incorporated into the master spec, but I think we should set that aside for now.

RS: You have a timeline. If there were certain things we wanted in the ARIA spec, if we get it in in time, would that meet your needs?

Tzviya: We'd like to avoid duplication (e.g. a chapter role in both the core spec and the DPUB spec).

RS: So would I.
... If we do it quickly, we could put it in a heartbeat draft. Then you wouldn't have to do it.

Tzviya: We'd want to know from the outset what direction we're headed in.

Janina: Trying to get all of this resolved in the W3C.
... PF is expected to take issues having to do with HTML to the task force.
... We are going to try to invite everyone soon, perhaps at next week's telecon, to discuss this.
... 11:00 Eastern.
... So we can discuss and move to a solution that works for everyone.
... We also need to discuss how a user agent developer knows how they have version x.y of a spec implemented.

RS: That's true for HTML now.

Janina: No, HTML extensions are part of the spec.

RS: I shared this ARIA extensions with Steve, and he doesn't seem to have any issues with it.

<Zakim> ShaneM, you wanted to talk about duplication and semantics

Shane: If something gets approved it's part of the core.
... If there are things that use the same term, obviously they need to be distinguished. Perhaps through a prefix.
... However, that's what PF is for.
... We need to ensure that roles work well across the board whenever possible.

RS: When a group creates a module, as long as we don't have a collision, we should have bottlenecks and trying to redefine industry language.
... These people working on this spec are experts in this area.
... We should try to be respectful and just worry about the conflict issues, and ensure we have good mappings.

CS: One issue is roles for an area. Another is OS features.
... These may need to be treated separately.
... From a user agent perspective, it's nice to be able to say "I've implemented this piece."
... For really big specs, that's more difficult.

<Zakim> MichaelC, you wanted to say it´s hard to predict future collisions, to decide what might need prefixing and what doesn´t

MC: About collisions, I think Shane is right. We're here to ensure there are not collisions.
... However that's hard to predict.
... Example if we were going to do music, "ARIA" might not have been chosen.

FE: Two things. One is you worry about what you can pass to ATs.
... With graphics, you want to expose that role to ATs, like "axis".
... You don't want to be limited to a small set of limited terms.
... The other thing is navigation. You need to provide keys and information so that you can adjust and provide better understanding.

<Zakim> tzviya, you wanted to ask about extensiblity of extensions

Tzviya: In Digital Publishing, we hope to be able to extend our extension in the future.
... For instance, terms related to education and assessment.

RS: That's why the nice thing about the prefix thing is that it will allow you to do what you want to do without name collisions.
... Cynthia, we haven't even looked at the API extensions, but that doesn't mean we can't change this down the road.
... I think HTML changed it a couple of times in the past.
... So if we need to do something, we can do it later on.
... Does anyone have any fundamental issues with what we have at the moment?

Janina: I don't.

CS: Please read it.

RS: (Reads it)

https://www.w3.org/WAI/PF/wiki/ARIAExtensions

The above text is what RS is reading from.

CS: That all sounds great for adding roles.
... But that doesn't address extensions I'd like to add.

MK: In response to what Fred said, it's my understanding that ATs don't have to use the rolenames explicitly. They can translate it however they want/need to.
... I don't think you should assume that the rolenames exposed will not be the rolenames presented by ATs.
... ATs are not obligated to use the rolename as stated in the spec.

RS: Right. We cannot dictate the AT presentation.

MK: Example "combo box" might be presented as "pick list" (hypothetically)

<richardschwerdtfeger> q/

JS: The accessibility APIs frequently have ways to do localization.

<Zakim> ShaneM, you wanted to point out that role synonyms are very low cost to support and to mention that collisions are not just about the LEXICAL name. They are about the behavior and

Shane: James Craig has stated that role synonyms are very low cost to support. Example, role presentation and role none.
... But, it's always been my assuption that the role which gets matched by the browser is exposed to the ATs.
... But I'd hope the other roles can still be obtained.
... As an AT. I can find out the real role value.

MK: I'm not sure that's the case on all platforms.

CS: In Windows, you can.
... We use the constant for the similar role, along with a localizable string that's closer to the ARIA role.
... I think Apple has a similar mechanism.

<clown> http://w3c.github.io/aria/core-aam/core-aam.html#roleMappingGeneralRules

JS: The URL above, says "User agents must expose the WAI-ARIA role string if the API supports a mechanism to do so. This allows assistive technologies to do their own additional processing of roles."

CS: The goal is to have the platform APIs be as rich as ARIA.

<Zakim> clown, you wanted to mention item 5 at http://w3c.github.io/aria/core-aam/core-aam.html#roleMappingGeneralRules

Shane: The other point I wanted to make is that it's not about the string, it's what the string means.

Janina: I was going to suggest we have some global statements and some module statements.
... I think this is what Cynthia was talking about earlier.
... Example, someone might want to extend states.
... We should probably add a statement that the extensions currently under consideration pertain to roles. But that there are potentially others.

Shane: What Cynthia is talking about is dynamically extending the taxonomy.

CS: (Confirms that)
... I heard several constraints about adding roles and properties, and not other features.
... We're going to need to deal with this.

RS: I think we'll have to deal with it for ARIA 2.
... For example, the WAPA work.

CS: It doesn't say that.

RS: Because I didn't think of it.
... Do you want us to say this for ARIA 1.x?

CS: I'll look at it later today. I don't want there to be an ARIA 1.2; I want to get started on ARIA 2.0.

<Zakim> janina, you wanted to propose generic vs specific delineation

RS: Whatever we do for 2.0 cannot break the 1.1 extensions.

Shane: Cynthia, please feel free to edit the wiki.

RS: Janina, can we add this to next Wednesday's call?

Janina: Yes.

RS: Aside from any additions by Cynthia, are we all set?
... Should we put a version on this?

CS: You may want to do so as I may have some significant edits.

Shane: I agree.

Review Open Action Items and Issues

action-1361

<trackbot> action-1361 -- James Nurthen to Suggest new text for the application role -- due 2015-04-02 -- OPEN

<trackbot> https://www.w3.org/WAI/PF/Group/track/actions/1361

JN: I haven't done it.

RS: I know. :)

JN: I'm not sure when I can get to this.

MK: I'm lined up to do aria-selected.
... If this action is important, I am willing to look at his action after I do aria-selected.

RS: How about I assign it to you Matt?

MK: Two weeks from today for action-1361.

<clown> action-603?

<trackbot> action-603 -- James Craig to Add thead markup to UAIG so headers are readable on multi-page printouts -- due 2010-06-30 -- CLOSED

<trackbot> https://www.w3.org/WAI/PF/Group/track/actions/603

issue-603

<trackbot> issue-603 -- Need an attr to indicate element activation triggers audio, video, etc. -- open

<trackbot> https://www.w3.org/WAI/PF/Group/track/issues/603

<clown> action-1363?

<trackbot> action-1363 -- James Craig to Patch issue-603: aria-startsmedia -- due 2015-03-12 -- OPEN

<trackbot> https://www.w3.org/WAI/PF/Group/track/actions/1363

action-1380

<trackbot> action-1380 -- James Craig to #presentation should mention aria-hidden vs presentation role on raster and vector images in relation to ACTION-1379 -- due 2015-02-13 -- OPEN

<trackbot> https://www.w3.org/WAI/PF/Group/track/actions/1380

RS: Can you look into this Joanie?

JD: I can look into it, but it won't be next week.
... June would be lovely.

JG: Testing plans for ARIA 1.1?

RS: We should probably do that.
... But should figure out what's going to happen with the charter.

MC: Testing is on my radar, but on the backburner currently.
... Do we want to use the existing harness or Test The Web Forward.
... The latter doesn't support direct access to accessibility APIs.

RS: The group would love that.

Janina: If we want to be in CR, we need to be testing.

CS: Six months might be doable.

RS: Should annotations be moved out to ARIA 2.0?

CS: Could we make it an extension?

RS: Why not move it out to ARIA 2.0. And if need be, put it into an extension. Will that work?

CS: Yes.

RS: I'm trying to get ARIA 1.1 done.

<richardschwerdtfeger> Meeting: W3C WAI-PF ARIA Caucus

<richardschwerdtfeger> chair: Rich

<MichaelC> http://www.w3.org/2015/04/draft-aria-charter

Charter

MC: http://www.w3.org/2015/04/draft-aria-charter
... The group should look at that.

- DRAFT -

Protocols and Formats Working Group Teleconference
14 May 2015

Attendees

Contents

Heads Up Items

Heartbeat

aria-interactive

ARIA Extension Review

Review Open Action Items and Issues

Charter

Summary of Action Items

Scribe.perl diagnostic output

- DRAFT -

Protocols and Formats Working Group Teleconference 14 May 2015

Attendees

Contents

Heads Up Items

Heartbeat

aria-interactive

ARIA Extension Review

Review Open Action Items and Issues

Charter

Summary of Action Items

Scribe.perl diagnostic output

Protocols and Formats Working Group Teleconference
14 May 2015