Protocols and Formats Working Group Teleconference -- 12 Feb 2015

<trackbot> Date: 12 February 2015

<asurkov> hi, clown

<clown> hi asurkov

<clown> welcome.

code 92473 not 2742?

<jamesn> I get the conference is restricted at this time

use 92473

<richardschwerdtfeger> https://lists.w3.org/Archives/Public/public-pfwg/2015Feb/0019.html

<mattking> zakim is telling me the conference is restricted

<mattking> i can not get in on phone

<jamesn> mattking - use 92473

rich I can scribe

<scribe> scribe: fesch

<richardschwerdtfeger> meeting: W3C WAI-PF ARIA Caucus

all - you can use 92473 to get in the call

aria-current

<clown> http://rawgit.com/w3c/aria/master/aria/aria.html#aria-current

<richardschwerdtfeger> - http://rawgit.com/w3c/aria/master/aria/aria.html#aria-current

mk: took an action for alternative text for a note - to better clarify current vs selected - not quite done yet

<richardschwerdtfeger> http://rawgit.com/w3c/aria/master/aria/aria.html#aria-current

rs: matt will put in updated note and we can discuss next week if needed

action-1561

<trackbot> action-1561 -- Richard Schwerdtfeger to Write up proposal for generic section role mappings to deal with div, role=“region”, and role=“group” confusion in the mappings -- due 2015-01-22 -- OPEN

<trackbot> https://www.w3.org/WAI/PF/Group/track/actions/1561

rs: needs Cynthia for 1561

Table Support

<clown> 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

<clown> action-1293

<trackbot> action-1293 -- Joanmarie Diggs to Add aria-colindex to role gridcell and aria-rowindex to role row -- due 2015-01-22 -- OPEN

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

jd: lots of discussion on row index and column index....

rs: what is the main issue?

mk: people don't agree on the main purpose of rowindex and columnindex

<clown> issue-398?

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

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

rs: describes use case of rowindex columnindex
... wants something similar to posinset and setsize

mk: colindex had a more confusing use case

rs: what was the confusion

<clown> https://lists.w3.org/Archives/Public/public-pfwg/2015Jan/0049.html

<clown> thread starts there ^

rs: not rowspan and colspan, so what appears to be missing is the number of rows and number of columns

jd: if I think of spread sheets, you can freeze and hide rows and columns
... believes this belongs on the cells - Alex doesn't
... believes colindex belongs on cells, didn't care about rowindex

<jamesn> cellindex should be on gridcell

mk: is this going to be defined as an attribute?

<jamesn> rowindex should be on row or gridcell

rs: will take an action to provide more detail, it matters where as it can bloat dowload

mk: doesn't see how ownership will work

jd: what if I want to mark up a data table, a colindex can only go on a cell

jn: what if you put it on the column header?

jd: Oh sure, that is fine

<jamesn> allow both on the cell but also allow rowindex on row and colindex on the columnheader for convenience

mk: it only works if you have a correct 1:1 mapping, so if someone does something where there is a multicell labeling structure, then you would make sure it gets assigned to the correct one
... if you put it on a 1:1 cells that will work, but if there are 2 numbers under a column - and you get these wonky structures

rs: you could put the colindex in the cell, then have it span cells, at the cell level
... would we do the same thing for rowindex

<jamesn> Janina - 92473

<janina> OK

rs: the important thing is the rowindex or colindex of the start - so if they span 3...

<clown> rawgit.com/w3c/aria/master/html-aam/html-aam.html#el-colspan

mk: in spereadsheets one dimension is numbered and one is lettered, so these are always numeric - or do the authors have control over number vs letter?
... does the value always have to be a number?

<asurkov> I think I like to have separate attributes for row/column spans

rs: if it was just a string might be a challenge

jd: potentially headers have a different properties, there are things that allow roman numerals and we can follow it

<clown> http://rawgit.com/w3c/aria/master/html-aam/html-aam.html#details-id-170

mk: sort of like a styling thing

jd: without deep diving.... (starts deep diving)...

jn: if they display on the screen they are similar to row headers

mk: they are different

rs: important to have a span and have them relate to the start, no notion of a column in HTML 5
... we have a rowindex which is the start of the row, and a colindex and max

jn: what does max mean?

<clown> aria-rowsize and aria-colsize?

<asurkov> maybe rowcount?

rs: like aria-setsize and aria-posinset

mk: do we want such a long attribute name - if you are putting on every cell

action-1293

<trackbot> action-1293 -- Joanmarie Diggs to Add aria-colindex to role gridcell and aria-rowindex to role row -- due 2015-01-22 -- OPEN

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

<clown> issue-398?

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

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

rs: taking action 1293

<clown> 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

jd: will turn the table stuff into a branch

jg: some elements proposed were eliminated

mk: count for last value, not max

Action-1362

<trackbot> Action-1362 -- James Craig to Patch issue-640: aria-hasmanagedfocus -- due 2015-01-29 -- OPEN

<trackbot> https://www.w3.org/WAI/PF/Group/track/actions/1362

handling of generic roles

action-1561

<trackbot> https://www.w3.org/WAI/PF/Group/track/actions/1561

<mattking> action-1561?

<trackbot> https://www.w3.org/WAI/PF/Group/track/actions/1561

<richardschwerdtfeger> https://lists.w3.org/Archives/Public/public-pfwg/2015Jan/0104.html

rs: has proposal - map a div to a section, just a container, we are overusing group
... list items use group on MacOS

CS: we don't have control types

rs: we need something generic

cs: we may need to accept group has become generic

mk: worry about existing implementations, but might fix some issues
... people use group where it makes noise

cs: big legacy of our mapping, narrator ignores group

rs: reads section of action

<clown> http://rawgit.com/w3c/aria/master/core-aam/core-aam.html#role-map-group

mk: reading group children

rs: when we use role=group now, where are we using it?
... are we using it for radio group?

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

mk: if you generize group, it won't affect it's children

<jamesn> it is a generic replacement for fieldset/legend too

rs: when you encounter group, do you expose group?

cs: IE exposes group, narrator ignores it
... a list is used a list - we treat lists and groups differently

too late

cs: we expose fieldset - which has a visual rendering

<clown> http://rawgit.com/w3c/aria/master/html-aam/html-aam.html#el-fieldset

cs: windows forms has them too

js: fieldset maps to group for everybody but ATK

cs: narrator does read groups in some cases, but it may have to deal with focusability ..
... willing to tweak behavior

rs: making generic takes care of legacy, but if you put it on a table, everything in the table comes out as groups

jd: a panel is a generic group of widgets

rs: want to be consistent, when you have a generic container with no label - want one mapping across implementations
... typically in SVG drawing attributes don't map, but if you make it significant - maybe by putting text on it... then we would like one consistent mapping between HTML and SVG

<clown> https://developer.gnome.org/atk/unstable/AtkObject.html#AtkRole

jd: we in ATK ... we solved xxxx by putting in role ATK_Static, worried that you will ruin stuff for us when we don't have a problem

mk: how does a div map in ATK?

chorus: role section

mk: want consistent role mapped for div and section

rs: what is confusing to me is panel and section

js: no they are different

mk: what we may want to come up with another aria role that maps to panel in ATK...

jd: if texty stuff maps to group and widgety stuff maps to panel I am happy

<clown> http://rawgit.com/w3c/aria/master/core-aam/core-aam.html#role-map-form

cs: on windows the pane role is a good one for widgety stuff

jd: your pane is our panel (NO PUNS)

<cyns> UIA pane from MSDN "The Pane control type is used to represent an object within a frame or document window. Users can navigate between pane controls and within the contents of the current pane, but cannot navigate between items in different panes. Thus, pane controls represent a level of grouping lower than windows or documents, but above individual controls. The user navigates between panes by pressing TAB, F6, or CTRL+TAB, depending on the context.

<cyns> https://msdn.microsoft.com/en-us/library/windows/desktop/ee671639(v=vs.85).aspx

mk: a problem with apple is both meaningful groups and non meaningful groups are mapped to groups

jd: apple is missing a subrole for list, may need a couple more subroles

rs: may need to create a panel role

mk: a lot of misuse of group is when we didn't have a better way
... need role for span

CS: when we were talking about annotations may have discussed it

<cyns> the network knew I was talking ;)

<richardschwerdtfeger> :-)

mk: can't finish this actionw without a mapping for span

<clown> http://rawgit.com/w3c/aria/master/html-aam/html-aam.html#el-span

<cyns> I think we may have covered the "span" thing when discussing annotations

mk: can't map span to same thing as div

jd: span needs to be updated for ATK

rs: do we agree we need some sort of panel role?

<cyns> generally, yes. Why call it panel an not pane?

mk: where does it go in the hieracrhy?

<richardschwerdtfeger> ACTION: joanie create a proposal for a panel role and modify the group role to be more generic [recorded in http://www.w3.org/2015/02/12-aria-minutes.html#action01]

<trackbot> Created ACTION-1581 - Create a proposal for a panel role and modify the group role to be more generic [on Joanmarie Diggs - due 2015-02-19].

<cyns> just to be platform-agnostic for a moment...

mk: is panel the right name for this?
... does tabpanel have children?
... should the new panel be a child of section and children of group move over to panel.
... rename group to panel.. and make a new role group

rs: can JD make a branch?

<LJWatson> http://bkardell.github.io/common-panel/

lw: can i mention the panel work?
... looking at a panel as a native HTML concept,

<clown> "On it's own, a panel may be used anywhere sectioning content is used, it is granted an aria-role="group"."

lw: have a single HTML architecture that can be dispalyed in multiple ways - which we currently use widgets for

<jamesn> sounds like a good idea

lw: a single panel seems weak
... important thing is getting a native HTML element

jd: lets work together...

jd & lw will coordinate

<clown> action-1362

<trackbot> action-1362 -- James Craig to Patch issue-640: aria-hasmanagedfocus -- due 2015-01-29 -- OPEN

<trackbot> https://www.w3.org/WAI/PF/Group/track/actions/1362

<clown> issue-640

<trackbot> issue-640 -- Investigate aria-hasmanagedfocus to indicate whether a region manages focus to enable complex widgets -- open

<trackbot> https://www.w3.org/WAI/PF/Group/track/issues/640

mk: wants to discuss aria-hasmangedfocus

rs: would like to make a list interactive and use arrow keys

mk: proposes a separate role
... big problem with an attribute when it gets to AT,

rs: why?

mk: may be ok if limited to a narrow group of roles ( list) which gives ability for list view and tree view

jn: lets not limit it

mk: all the name is saying that there is managed interaction on a static structure.

jn: using role of application is hacky,

mk: roles communicate interaction and

bg: is that like a modality?

mk: application does not give the user any idea what the interaction might be
... hasmangedfocus is a solution looking for a problem

jn: we have a ton of places where we need this

mk: maybe we need additional application roles, but application role is problematic for AT as it doesn't tell you about interactiviyt
... a toolbar should have multiple inheritance,

rs: list normally does not have arrow key navigation, you may want that

mk: should solve that with a role that inherits from widget
... if we have widget role, it should not map to a static element

jn: there are cases where we don't have the appropriate role - like a widget role, or listbox, which is hacky and cheesy

bg: has use cases where this important, one case where there is no role for columns
... problems with grids as well
... if need a modality of different widget types, that might be a valuable use case
... may have a widget that does two things at one time.
... task was able to make it accessible
... solved using role=application

mk: as a general principle that we would take a semantic of documents and turn them into a role-less widget. leave HTML semantics out of it.

- DRAFT -

Protocols and Formats Working Group Teleconference
12 Feb 2015

Attendees

Contents

aria-current

Table Support

handling of generic roles

Summary of Action Items

Scribe.perl diagnostic output

- DRAFT -

Protocols and Formats Working Group Teleconference 12 Feb 2015

Attendees

Contents

aria-current

Table Support

handling of generic roles

Summary of Action Items

Scribe.perl diagnostic output

Protocols and Formats Working Group Teleconference
12 Feb 2015