<trackbot> Date: 05 May 2014

<richardschwerdtfeger> meeting: W3C WAI-PF ARIA Caucus

<richardschwerdtfeger> chair: Rich

<chrisgallello> It might be me - I'm joining in to discuss annotations within ARIA

<richardschwerdtfeger> http://lists.w3.org/Archives/Public/public-pfwg/2014May/0014.html

<scribe> scribenick: clown

ISSUE-655 Consider annotation roles

<richardschwerdtfeger> ttps://www.w3.org/WAI/PF/Group/track/issues/655

CG: I'm on the office online team
... The problem:

<richardschwerdtfeger> https://www.w3.org/WAI/PF/Group/track/issues/655

CG: There are things like comments, footnotes, spelling errors, etc. that we consider to be annotations.
... We need to mark them up so that a SR can handle them.
... If you are familiar with MS Word, comments work much the same way with office online.
... It renders as ancillary text, and I sighted user can decide whether to read it.
... Would like similar behaviour in the case of ATs.
... There is a link to a position paper in the issue that describes this a bit more.
... In particular, extensions to aria.
... Three new aria tags.
... Aria-annonationtype — a localized name of the annotation type.
... For example, for a comment, there is the comment text, and then around it is <?>

<richardschwerdtfeger> http://www.w3.org/WAI/PF/aria-1.1/states_and_properties#aria-invalid

JC: The localized annotation name, is something like the localized role name we are currently discussing.

CG: where can I catch up on the work being done on that?
... That sounds like a way to implement this.

RS: We actually use the "grammar" and "spelling" for each of the text elements.
... It was working fine, but you are seeing limitations?

CG: Yes, but we should check it again.

RS: A problem is that some AT vendors have not modified their code given the recent updates.

CG: The other two proposed tags also mark up the connection the text and its annotation.
... You can think of these tags like aria-describedby.
... aria-annotatedby would identify the comment div on the page.
... And, in order to go backwards, there is aria-annotatedfor to list the IDs.

BG: Is this like aria-flowto?

CG: Flowto is more for "go read this next".
... This is more as a way of calling out that there is an annotaion here.

JN: We use describedby for lists, but it's not perfect.
... To cover excel to cover the extra info that can be stored on the cell.

MZ: I wanted to comment on the annotedby and annotedfor.
... Does it need to be both ways?
... FF creates the reverse relation for things like describedby.

<jcraig> ack +49

MZ: Can't we make the job on authors easiers.

RS: I think we should do just one.

JC: I prefere the "by" and not the "for".

MZ: I spoke to some people in the Windows team about flowsto and they wanted to create a flowsfrom
... And they found it easier if there was only one direction.

JC: Did they say what the problem is if the reverse does not match up?

MZ: No.

JC: That is part of the discussion. To get references in sync.
... It's easier if it's just one direction.

<jamesn> +1 to that

RS: Let's summarize where we have agreement. Is "by" preferred to "from"?

<someone> : agreed.

JC: I had some questions on the examples.

CG: Which example?

JC: The one called basic example.
... There are no roles in either scenario.
... The first one makes sesnse since it is just text.
... But the the other one might need a role, with a description of it being a deletion.
... The screen reader could detect the annotation type by a refereence to the annotation.

CG: There are times with annotation where the text is not in the DOM.

JC: Do you have an example.

CG: If the comments pane is not shown, we remove it from the DOM.
... Perhaps we should have it in the DOM, but marked hidden.

RS: That makes sense.

JC: Do we need an IndieUI action to ?
... To indicate that the user wants to see this now.

CG: That fits with my thoughts do we need to create something else for hidden annotations.

<Zakim> jamesn, you wanted to say we have that

CG: Is there something similar to an invoke action?

JC: There is something similar going on in the IndieUI WG.
... Does a click on a link activate the link? Or the annotation associated with it?

CG: Clicking the link opens up the new page.

JC: Is it like a contextual menu?

<Zakim> jcraig, you wanted to say we should be careful not to conflate the API with UI

JC: That might be sufficiently covered.
... We need to not confuse the UI with the API we are developing.
... We may be situations where the user wants to hear the entirety of an addition, but only just that things have been deletions.
... But we want the API supports all of these scenerios.

MZ: I agree with that, and it could be controlled by the screen reader's verbosity settings.
... It's the user's choice or the ATs how to present it.
... But we want the information on hand to decide that.

RS: right.

MZ: I think the filtering should be done by the screen reader, not the web app.

JC: But, in the case that Chris is describing, there may be reasons why the comments are not in the DOM — performance reasons.

CG: what you bring up about lazy loading — that affects the web app in general.
... And we need to account for lazy loading in general.

RS: we also have to think about validation tools.
... If you have an annotation reference, you have to deal with that issues.

JC: If there's an idref but it's not yet loaded, you can't say that's an error.

RS: Is busy involved here somehow?
... We can have something in the annotation that says it's not loaded, and that might handle it.
... And, also, let the user load the comment on demand.

JC: Marco's proposal to have all the notes in the DOM makes things easier for the UA and that AT.
... But, Chris, you should take these thoughts back and see where performance in a real issue.

CG: If we load the comment pane, but mark it with aria-hidden, would that be a solution?

JC: That would do.

JN: We need to couple the situation where the info is not there, until it is requested and then see how the load affects things.

RS: Worry: why have the SR user have access to the comments but the sighted user not?

CG: We can have a gesture where the sighted user opens the comment pane to view it.

MZ: The reason we did this in JAWS because the sighted user could use mouse hove to get the comment.

CG: Let's leave this open, and I'll take this back to the team, and can we discuss on list.

JC: Keep the discussion on the list, please.
... In the discussion topic section, there is noted annotation types: comment, spelling error, etc.
... These seem something like we should tokenize and include as roles or annotationtype as an enumerated type rather than a localized string.
... Because of the scenario where the annoation is not in the DOM, then it's harder

<cyns> been trying to get though on the phone, and conf is full. anything we can do abt that?

JC: The localization can be done by the user agent or the AT.

CG: we don't know what the other types are going to be.
... My concern that a tokenized list is too constrictive.

JC: But, the problem is there are different scenarios such as where the user wants to hear all the comments.

CG: Are you suggesting different types of annotation?

JC: Partly, but a screen reader can't tell from the localized string what the type is.
... But a token would solve that.

<Zakim> clown, you wanted to note that cyns cannot get on the call.

LJW: That's very important from a user's perspective to be able to turn different types of comments on and off.

RS: Chris, are you going to take this back to the team, and revisit next week?

CG: Sure.
... One more issue.
... If you have overlapping annotations like a comment and an enclosed footnote.
... Or just two comments. How is that marked up?
... So there is a programmatic way to note that the overlap and how they both are associated with the same span.
... We could discuss this on the list.
... <description of how it is done now>
... It's all written up in the document.

<Zakim> jcraig, you wanted to suggest vendor-prefixed token values and to

JC: <describes how it might be done programmatically>

CG: You might need two annotatedbys and which goes with which.
... Is the suggestion that annotatedby has a list of IDREFs?
... What is described in the doc is not for overlapping annotations.
... But that is something to consider.

CG/JC: <more discussion of how overlapping comments might be handled>

CG: a concern is if there are say 3 different annotations on a span, there should be a way to jump to the different types, say the spelling error.

JC: And that keyboard shortcut would depend on?

CG: If the screen reader can recognize error and comment differently, then you can have different ways to get at them.

JC: The only problem if there are two comments about the same bit of text, or the same annotation type.

CG: The sighted user could tell by visual cues.
... I like these thoughts, and we can contiue on list.

RS: When we have a design we can meet with the AT vendors.

CG: maybe we should revisit this in two weeks time?

<Zakim> jcraig, you wanted to suggest vendor-prefixed token values and to

<jgunder_> Telephone bridge is full, won't let me on

JC: Regarding extensibility of tokens, some enumerations are set up to be extensible.
... See aria-invalid. Besides the specified tokens, any other string is regarded as "true".
... And we could include vendor prefixed tokens.
... Something aria-invalid="MS:notelink".
... We could do something similar for the annotation types.

MZ: I like the conccpt.

JN: I don't. Vendors means anyone? Or ATs in the field?

JC: It's a slippery slope.
... It's technically valid now.
... The worry is AT vendors will lock into these extra tokens.

JN: Why not put the tokens in ARIA 1.1. Why not solve the problem?
... Say that this is most like the other ones, rather than just "true".
... Just like we have with roles: use the first recognized one.
... aria-invalid is weird since you can state that something is invalid without saying what the error is.
... Let's not make the same mistake with annotations.

RS: Let's wrap, and plan on having Chris back in a week or two.

: Meeting time.

RS: Marco, we got a number of west coasters to attend given this meeting time.
... So, it's difficult to move the meeting time.

MZ: My conflict is the Mozilla product wide meeting, starting at 1:30 pm EDT.
... And it lasts as long as there are topics.

JC: We can arrange that your topics are at the beginning of this call.

MZ: That will help.

RS: Difficult to reschedule.
... Send me a note at what you want discussed, and we will do it first.

ISSUE-606

issue-606?

<trackbot> issue-606 -- grid role does not allow aria-setsize or aria-posinset -- open

<trackbot> https://www.w3.org/WAI/PF/Group/track/issues/606

RS: what is the use case, James?

JC: This might have been brought up by Dominic.

RS: You could have multiple grids in a row.

JC: That doesn't not sound right.
... I think this was on the descendant nodes.

JN: rows and columns would be handy, but I thought that was ARIA 2.0

RS: we have rows now.

<jcraig> https://dvcs.w3.org/hg/pfwg/raw-file/tip/ARIA/1.0/tests/test-files/roles-properties-supported/roles-properties-supported-grid-aria-level-1.html

JN: we could put it on cells.

RS: This is the inherited.

JC: I think this is an invalid issues, and a bad example.
... Remove it from the test suite.

RS: This was specifically for a grid. We should be correct the example.
... I can see the value of putting it on rows and cells.

JC: We should make an action for someone to remove or fix the test file.

RS: Michael, could you correct this one in the test suite?

<jcraig> From the validator:

<jcraig> "https://dvcs.w3.org/hg/pfwg/raw-file/tip/ARIA/1.0/tests/test-files/roles-properties-supported/roles-properties-supported-grid-aria-level-1.html":8:83: error: Attribute “aria-setsize” not allowed on element “div” in this context.

<jcraig> "https://dvcs.w3.org/hg/pfwg/raw-file/tip/ARIA/1.0/tests/test-files/roles-properties-supported/roles-properties-supported-grid-aria-level-1.html":8:83: error: Attribute “aria-posinset” not allowed on element “div” in this context.

JC: Here is the validator error.

<jcraig> action to fix examples

<trackbot> Error finding 'to'. You can review and register nicknames at <https://www.w3.org/WAI/PF/Group/track/users>.

RS/JC: We want someone to look at the test cases and see if the tests are valid, and fix as necessary.

MC: That I can do with a query.

RS: I'm looking at the row role, and there is no posinset on it.

http://www.w3.org/WAI/PF/aria-1.1/states_and_properties#aria-setsize

<MichaelC> Testcase 178

MC: I see only one test file, James. Are there more?

JC: Possibly...

<richardschwerdtfeger> http://www.w3.org/WAI/PF/aria/roles#row

<richardschwerdtfeger> http://www.w3.org/WAI/PF/aria/roles#gridcell

JC: Check test cases on either side, perhaps.
... I think this is the real issue, but referencing the test case was not correct.
... If it's on row, it's vertical, and on gridcell it's horizontal. It's like rowindex and colindex.
... I'm not sure we want to add this to row and columns.

JN: it's a problem when it's both horizontal and vertical.

RS: So, can you take an action to fix the test cases, Michael?

MC: I don't know what the fix is.

JC: And the fix depends on the disposition of these other issues.

RS: Do we need this now?

<jcraig> related to issue-398

<jcraig> issue-398?

<trackbot> issue-398 -- Errata: aria-setsize and aria-posinset should apply to more than just listitem and option -- open

<trackbot> https://www.w3.org/WAI/PF/Group/track/issues/398

JN: We end up not using gridcell since it doesn't work, and we end up using labels to make things work.

<jcraig> action-1293?

<trackbot> action-1293 -- James Craig to Add aria-colindex to role gridcell and aria-rowindex to role row -- due 2013-11-11 -- OPEN

<trackbot> https://www.w3.org/WAI/PF/Group/track/actions/1293

JC: I do have an action to add colindex and row index.
... We could bump that to 1.1

RS: We could be rowindex on the row, and the cell inherit it.

JN: How would these indices work?

JC: It disambiguates the meaning of setsize and poinset

JN: I guess it helps, but I'm not sure it's necessary.
... It only disambiguates one.

JC: It diambiguates both since it's rowindex and colindex.

<jcraig> Dominic Mazzoni, "I'd prefer if we added aria-colindex and aria-rowindex attributes, rather than reuse aria-posinset to mean rowindex when there's no equivalent for col index."

JN: Okay, I'm good.

RS: Would it help if moved to 1.1?

JN: It would help.

<jcraig> action-1309?

<trackbot> action-1309 -- Cynthia Shelly to Determine the best place to map aria-colindex and aria-rowindex for msaa+uiexpress and uia. -- due 2014-01-31 -- OPEN

<trackbot> https://www.w3.org/WAI/PF/Group/track/actions/1309

JC: Cynthia, you also have an action.

CS: I can take a look at it.
... What is the due date?

RS: Colindex was a due date of November 2013.
... Do you want to change it, James?
... For a heartbeat draft?

JC: I want as much in the hearbeat draft as possible.

RS: Can we make this a 1.1.?

JC: Done.

<jamesn> thanks

RS: Can you look for other holes, James N?

JN: Yes.

JC: Can we close the issue?
... We can close 398 and 606
... I can update the example as I make the spec changes.

RS: Anything else?

This is scribe.perl Revision: 1.138  of Date: 2013-04-25 13:59:11  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: s/aria-desribedby/aria-describedby/
Succeeded: s/on column/on gridcell/
Succeeded: s/row index/rowindex/
Found ScribeNick: clown
Inferring Scribes: clown
Default Present: Rich_Schwerdtfeger, Cooper, Joseph_Scheuhammer, janina, Bryan_Garaventa, LJWatson, Suzann_Keohane, jcraig, Chris_Gallello
Present: Rich_Schwerdtfeger Cooper Joseph_Scheuhammer janina Bryan_Garaventa LJWatson Suzann_Keohane jcraig Chris_Gallello
Agenda: http://lists.w3.org/Archives/Public/public-pfwg/2014May/0014.html
Found Date: 05 May 2014
Guessing minutes URL: http://www.w3.org/2014/05/05-aria-minutes.html
People with action items: 

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