Minutes March 19, 2015 WAI-PF ARIA Caucus

https://www.w3.org/2015/03/19-aria-minutes.html


  Protocols and Formats Working Group Teleconference
  19 Mar 2015

Agenda <https://lists.w3.org/Archives/Public/public-pfwg/2015Mar/0069.html>

See also: IRC log <http://www.w3.org/2015/03/19-aria-irc>


    Attendees

Present
    fesch, Jon_Gunderson, Joanmarie_Diggs, Michael_Cooper, Janina,
    Rich_Schwerdtfeger, Joseph_Scheuhammer, James_Craig,
    Bryan_Garaventa, Matt_King, Cynthia_Shelly
Regrets
Chair
    Rich
Scribe
    MichaelC


    Contents

  * Topics <https://www.w3.org/2015/03/19-aria-minutes.html#agenda>
     1. HTML AAM <https://www.w3.org/2015/03/19-aria-minutes.html#item01>
     2. Action 1581 Create a proposal for a panel role
        <https://www.w3.org/2015/03/19-aria-minutes.html#item02>
     3. Table grid related actions
        <https://www.w3.org/2015/03/19-aria-minutes.html#item03>
     4. Action 1349 Patch issue-561: we need @aria-placeholder
        <https://www.w3.org/2015/03/19-aria-minutes.html#item04>
  * Summary of Action Items
    <https://www.w3.org/2015/03/19-aria-minutes.html#ActionSummary>

------------------------------------------------------------------------

<trackbot> Date: 19 March 2015

<richardschwerdtfeger> meeting: W3C WAI-PF ARIA Caucus

<richardschwerdtfeger> 
https://lists.w3.org/Archives/Public/public-pfwg/2015Mar/0069.html

<scribe> scribe: MichaelC


      HTML AAM

followed up with HTML A11Y TF from our discussion last week

identified many points of documented joint dependency, so we´re ok on 
that front

need to figure out who will stay on top of filing bugs

ack s/followed/js: followed/

rs: we can take turns

I don´t usually look at FPWD

but for this one, we need to both here in ARIA, and in the AAPI group

js: ARIA in HTML

rs: for HTML to exit CR, will they be required to meet the mapping guide?

<janina> 
http://rawgit.com/webspecs/html-aria/master/index.src.html?specStatus=FPWD;publishDate=2015-03-11;wg=HTML%20Working%20Group;wgURI=http://www.w3.org/html/wg/;wgPublicList=publi

js: not sure if that´s established

<richardschwerdtfeger> 
https://specs.webplatform.org/html-aria/webspecs/master/

mc: for ARIA, we used mappings to define testability of the spec

we could offer the mappings as a way of testing part of the HTML spec

don´t know if we would say it must be tested in and of itself

<richardschwerdtfeger> 
http://rawgit.com/webspecs/html-aria/master/index.src.html?specStatus=FPWD;publishDate=2015-03-11;wg=HTML%20Working%20Group;wgURI=http://www.w3.org/html/wg/;wgPublicList=public-html

<scrambling about for the right URI>

js: HTML intends to retire editors´ drafts

instead, publish daily to TR

rs: still, let´s use the above rawgit

clown: is that URI frozen?

mc: no, unless there is a specific commit in the URI, that can change at 
any time

note that´s not a WD, it´s just formatted as if it were

clown: think a CfC should be done on a branch that won´t have 
non-editorial changes

mc: I usually do that, but I don´t have hands on every CfC

we could / should make this our procedure across the board

could also ask HTML to do so, don´t know if they want to

<richardschwerdtfeger> *ACTION:* Joseph Review and provide feedback to 
Steve on ARIA in HTML [recorded in 
http://www.w3.org/2015/03/19-aria-minutes.html#action01]

<trackbot> Created ACTION-1598 - Review and provide feedback to steve on 
aria in html [on Joseph Scheuhammer - due 2015-03-26].

http://rawgit.com/webspecs/html-aria/d9d5547ff647268ad6ab2497eb5fa443b2032f31/index.src.html?specStatus=FPWD;publishDate=2015-03-11;wg=HTML%20Working%20Group;wgURI=http://www.w3.org/html/wg/;wgPublicList=public-html

<^ frozen version of the same doc>

action-1598: ARIA in HTML 
http://rawgit.com/webspecs/html-aria/d9d5547ff647268ad6ab2497eb5fa443b2032f31/index.src.html?specStatus=FPWD;publishDate=2015-03-11;wg=HTML%20Working%20Group;wgURI=http://www.w3.org/html/wg/;wgPublicList=public-html

<trackbot> Notes added to action-1598 Review and provide feedback to 
steve on aria in html.

<where to file bugs?>

jc: W3C bug tracker, HTML product, some component, should auto assign to 
Steve

<joanie> 
https://www.w3.org/Bugs/Public/buglist.cgi?component=Using%20ARIA%20in%20HTML&list_id=53494&product=HTML%20WG&resolution=---

<struggles to find component, looking at 
https://www.w3.org/Bugs/Public/enter_bug.cgi?product=HTML%20WG 
<https://www.w3.org/Bugs/Public/enter_bug.cgi?product=HTML%20WG>>

<clown> action-1598?

<trackbot> action-1598 -- Joseph Scheuhammer to Review and provide 
feedback to steve on aria in html -- due 2015-05-21 -- OPEN

<trackbot> https://www.w3.org/WAI/PF/Group/track/actions/1598

js: confused, the document referenced by the URI above is a different 
document than the one we´re considering

<richardschwerdtfeger> https://github.com/webspecs/html-aria/issues

rs: Steve wants bugs in GitHub issues ^

jg: should this include only ARIA 1.0, or also ARIA 1.1?

js: ARIA 1.1, even though the doc itself is a 1.0

<richardschwerdtfeger> 
https://lists.w3.org/Archives/Public/public-pfwg/2015Mar/0069.html


      Action 1581 Create a proposal for a panel role

jd: went down a rabbit hole

have more questions than answers

but working on it

have posed questions by email, can discuss here if people have thoughts

gaps, user agent normative statements, etc.

some of this is straightforward, other more meaty


      Table grid related actions

action-1395?

<trackbot> action-1395 -- Joanmarie Diggs to Add table role to taxonomy 
and aria 1.1 spec. -- due 2015-02-12 -- OPEN

<trackbot> https://www.w3.org/WAI/PF/Group/track/actions/1395

action-1293?

<trackbot> action-1293 -- Richard Schwerdtfeger to Add aria-colindex to 
role gridcell and aria-rowindex to role row -- due 2015-02-26 -- OPEN

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

<richardschwerdtfeger> 
https://lists.w3.org/Archives/Public/public-pfwg/2015Mar/0068.html

action-1403?

<trackbot> action-1403 -- Joanmarie Diggs to Add table/cell roles, 
rowspan/colspan attrs, related taxonomy changes, and suggestions for 
html mapping document -- due 2015-02-13 -- OPEN

<trackbot> https://www.w3.org/WAI/PF/Group/track/actions/1403

jd: working on that stuff

have been having list discussion

everyone agreed should be rowcount, rowindex, etc., instead of rcount, 
rindex

rs: people want to type all that extra?

jc: mostly won´t be typed by humans

where there is manual typing, what we lose in brevity we gain in clarity

rs: worth the extra bytes?

could get large

jc: if we really wanted to minimize download we´d go to rc and ri and be 
even more obfusticated

jd: don´t think it´s as large as it seems

rs: was just trying to save file size, but ok if you all prefer this

I have gotten feedback about overly long attributes

jd: just 2 extra characters

I went through many mailing lists to determine a consensus

jd: Dominic said there are cases where aria-rowcount needs to be on each 
cell

I asked what those use cases are

I think it should be on the table

same for colcount

mk: these counts are so you don´t have to have entire table in DOM?

they never change?

clown: could change if the underlying data set changes on the server

jc: treegrids with expanded rows

DOM parents could be 2 levels up

not always available

we could make these optionally on the container

but need to make available for the edge case

mk: edge case doesn´t require at cell level, but at row level

jc: there could be a pivot table

in which it´s on the other axis

jd: say whole table is visible, but only some on screen

if it has 200 rows, but 50 on screen

rowcount is 200?

rs: yes

jd: why does that need to be on each row?

rs: do you want to make AT go up to top of table?

jd: the mapping spec could decree that it be exposed as needed

but why does *author* need to put it in 200 times?

rs: one major browser requires AT to go into DOM

jc: for me the reason is, say you have a very large data set with 
collapsed rows etc.

mk: @@

cs: IE6 is also an edge case ;)

mk: @@

jc: usually you wouldn´t have to do that
... not too expensive to walk up a single ancestor chain

cs: but it´s less expensive than @@

mk: @@

jd: so for now, putting it on the table / container that holds the rows

jc: either table or rowgroup

rs: IE users might not benefit from that

cs: we don´t need to optimize for an unsupported 15 year old browser

<clown> <div role="table" aria-rowcount="5000"> or <div role="rowgroup" 
aria-rowcount="5000">

rs: IE 11 will still ship

AT vendors will need to rewrite their code

cs: as we´ve been asking for 10 years

rs: with our clients, some have just now moved to IE 8

they demand that our products work with the old tech

they don´t care about the shiny new thing

mk: so put on table with option of row

allows authors to do that if they decide they need to

cs: <talked over again>

js: trying to put a plan for authors that will work

we need a sit-down with the AT developers

cs: yes

we´re working on that

jc: sometimes a stake in the ground is needed

cs: we´ve been warning them 10 years they need to catch up

rs: let´s see what we can do to get them moving

I´ll do whatever it takes

meanwhile, seems like we have a vehicle to make them not look so far

so JD, let´s put in both places

jd: will do

same question applies to colcount, will take similar approach

though different for cols because they don´t have same parent container 
as rows

jd: another question - does guidance about handling rowcount belong in 
ARIA spec or mappings?

rs: put in ARIA so we wouldn´t forget about it

jd: think Joseph and I will be sure to not forget :)

rs: author will probably not look at mappings spec

<clown> http://w3c.github.io/aria/aria/aria.html#aria-setsize

for setsize etc. we mention it´s automatically computed

jd: yes, but that´s because author can omit if the computation will work

but for rowcount, you want UA to apply to all descendant cells?

my proposed wording is ¨expose in way platform expects it¨

which varies by platform, not always defined on cells

think this is mapping detail, not ARIA

rs: that was placeholder to be sure it was covered, since I wasn´t 
addressing mapping spec separately

do think @@ needs to stay there

jd: thinks I´ve preserved that

<clown> If all items in a set are present in the document structure, it 
is not necessary to set this property, as the 
http://w3c.github.io/aria/aria/aria.html#dfn-user-agent can 
automatically calculate the set size and position for each item.

clown: ^ for aria-setsize

rs: if we have something like that, it´s fine

<joanie> If all of the columns are present in the document structure, it 
is not necessary to set this attribute as the user agent can 
automatically calculate the column index of each cell or gridcell. 
However, if only a portion of the columns is present in the document 
structure at a given moment, this property is needed to provide an 
explicit indication of the column of each cell or gridcell with respect 
to the full table.

jd: ^ is what I did

took out some UA MUST from the ARIA spec

rs: how about you put all this in a branch for us to review

jd: <apoplectic>

many questions being answered now

<clown> If all of the columns are present in the DOM, it is not 
necessary to set this attribute as the user agent can automatically 
calculate the column index of each cell or gridcell. However, if only a 
portion of the columns is present in the document structure at a given 
moment, this property is needed to provide an explicit indication of the 
column index of each cell or gridcell with respect to the full table.

hadn´t done until now because of too many open questions

jd: another was, could colindex be implied, not required on every cell?

if there are no gaps, it can just increment for each col

but what are situations where there can be gaps?

so what are situations where you could leave index out?

mk: say some columns are hidden

rs: do you skip over them?

mk: like you see in Excel

if you scroll way right, header column may still display but ones in 
between could be hidden

jd: posted example of hiding a col in Google docs

clown asked if it´s in DOM

was looking, can´t seem to find the right part of the DOM to look

so still open question, do we allow gaps or not?

do authors have option of not writing all colindices, or not?

rs: hidden presumably mean can be expanded?

jd: of course

clown: I wasn´t thinking about hidden cols when I brought up question on 
mailing list

I was thinking about when selected rows are put in the DOM for some reason

for setsize etc. that wasn´t a consideration

but is this a situation?

rs: if they´re hidden but in DOM, it´s different from not loading into DOM

hidden should mean user can choose to expand

<joanie> <div role="row">

<joanie> <span role="gridcell" aria-colindex="3">Jane</span>

<joanie> <span role="gridcell" aria-colindex="4">Jones</span>

<joanie> <span role="gridcell" aria-colindex="5">Acme, Inc.</span>

<joanie> <span role="gridcell" aria-colindex="9">555-1234</span>

<joanie> </div>

jd: ^ is that valid, or not?

cols 6, 7, 8 are hidden

clown: then they´re in the DOM

jd: so that would never happen

clown: that was my question, can this happen? can webapp get specific 
columns, or must there always be a contiguous set?

jd: if they´re not there because author has chosen to hide

clown: in past, if you picked a subset, you still always got a 
contiguous subset

do we continue that, or anticipate that non-contiguous set?

rs: web author knows more about their app than we do

so they should override

clown: in which case colindex required on every cell

haven´t anticipated with setsize, posinset

we might need to tweak those also

mc: require on all cells, or just allow for this use case?

clown: if set is contiguous, only needed on row

but if non-contiguous, need on every cell

same for rowindex

jd: ok

now on to rowspan / colspan

could have different colindex in different rows because of spanning

does author need to put colindex everywhere here too?

I really think colindex needed on each cell, for reasons like this

rs: unless it´s computed

clown: @@

jd: with tables, if things broken in code, table visually looks wrong

clown: if author can get it right, allow them to put the indices where 
it´s implied

but otherwise put everywhere

jd: but they´ll all think they did it right, when they didn´t

so I propose we start with ¨authors must provide everywhere¨ and see if 
complaints raised during wide review of spec

mk: if author chooses to put index value on one cell on a row, they must 
put it on all cells in the row; there is no implied incrementing; if 
they put it on the row, it can be implied (propagated by UA for all 
cells in the row); if they put it on one row, they have to put it on all 
rows

jd: brilliant

I´ll write that down

rs: if you have colindex with colspan, colindex is start of colspan

jd: plan to be very explicit about that stuff

rs: want to minimize author MUST

clown: JG can put in conformance checking tools

jg: sure

jd: some examples have HTML tables as ARIA grids

others divs as ARIA grids

with divs you have to be pretty explicit with your aria

but with tables, it should be ok to omit certain ARIA properties if the 
native HTML one is correct

jc: yes, and you´re supposed to use native semantics where available and 
not override with ARIA

definitely applies here

jd: ok, now have enough to put together a proposal for wordsmithing


      Action 1349 Patch issue-561: we need @aria-placeholder

action-1349?

<trackbot> action-1349 -- Joanmarie Diggs to Patch issue-561: we need 
@aria-placeholder as backup for @placeholder in custom fields. -- due 
2015-03-05 -- OPEN

<trackbot> https://www.w3.org/WAI/PF/Group/track/actions/1349

jd: awaiting reply to my proposal

<richardschwerdtfeger> https://www.w3.org/WAI/PF/Group/track/actions/1349

<clown> 
http://rawgit.com/w3c/aria/action-1349/aria/aria.html#aria-placeholder

rs: maybe a different thing for the value

jc: nearly used in input textarea and contenteditable

using custom placeholders because they want different visual behavior

e.g., a div that hovers over top

that looks similar, but behaves differently for AT user

not sure we need value back in

most of the time, aria-placeholder would be used instead of @placeholder

do need to make sure ARIA version allowed to override

jd: is this just a terminology confusion since ¨value¨ has specific meaning?

rs: something about control contains no text, or content...

agree we need to be able to override

clown: you shouldn´t have both

rs: but if you do, ARIA needs to win

mk: don´t we usually specify the opposite?

e.g., for tables

clown: on table, if you have rowspan and aria-rowspan, rowspan wins

here, we say aria-placeholder wins over placeholder

jc: this is like aria-label overriding @alt

think there is action to add placeholder computation to the text computation

so this isn´t just a regular attribute, it´s part of computed value

rs: so we need to bring this to Steve´s attention

jd: ok, I´m gonna start using ¨text for content¨ instead of ¨value¨

should we add notes about what wins, or author shouldn´t do both?

if author using native, they shouldn´t use the ARIA one?

jg: what does placeholder do?

clown: hint as to what author should provide

jg: do you still also need aria-label?

various: yes

jg: what if it´s the same

clown: that would be dumb

jg: but that´s a common pattern

so, intent of aria-placeholder is a hint, not labeling

but in practice, people mix those up

jd: so, do we need to clarify native vs ARIA?

clown: we do elsewhere, such as aria-label

jd: ok, will look at that

clown: so need to define AAPI mappings?

would like to define a ¨placeholder¨ property and let AT sort

is it the browser´s job to put placeholder in the name, or AT´s job to 
fetch it?

rs: if you have text box with title and placeholder, which wins?

jd: title

<richardschwerdtfeger> The placeholder attribute represents a short hint 
(a word or short phrase) intended to aid the user with data entry when 
the control has no value. A hint could be a sample value or a brief 
description of the expected format. The attribute, if specified, must 
have a value that contains no "LF" (U+000A) or "CR" (U+000D) characters.

clown: HTML says use placeholder as last resort


    Summary of Action Items

*[NEW]* *ACTION:* Joseph Review and provide feedback to Steve on ARIA in 
HTML [recorded in http://www.w3.org/2015/03/19-aria-minutes.html#action01]

[End of minutes]
------------------------------------------------------------------------
Minutes formatted by David Booth's scribe.perl 
<http://dev.w3.org/cvsweb/%7Echeckout%7E/2002/scribe/scribedoc.htm> 
version 1.140 (CVS log <http://dev.w3.org/cvsweb/2002/scribe/>)
$Date: 2015-03-19 18:02:16 $

------------------------------------------------------------------------