513848 - nsHTMLTableCellAccessible is used in dojo's crazy ARIA grid

Status: RESOLVED FIXED

(Core :: Disability Access APIs, defect)

Description

[alexander :surkov (:asurkov)]

Simple example would be

<div role="grid">
  <div role="row">
    <table role="presentation">
      <tr>
        <td role="gridcell">cell1</td><td role="gridcell">cell1</td>
      </tr>
    </table>
  </div>
</div>

Comment 1

[David Bolter [:davidb] (NeedInfo me for attention)]

Is the problem here that we are not respecting the role="presentation"?

Comment 2

[alexander :surkov (:asurkov)]

Sort of, it depends on how to treat your question :) accesibles have ARIA role so we don't test table structure at all and tests failed when we decide if we should use ARIA table specific accessible classes.

Comment 3

[David Bolter [:davidb] (NeedInfo me for attention)]

Can you explain the bug in a little more detail? :)

Comment 4

[alexander :surkov (:asurkov)]

We should create nsARIAGridCellAccessible for HTML td elements under HTML table which is presentational because all HTML table accessible classes are intendent to work with HTML table frame methods. And this doesn't make sense because obviously HTML table with role="presentation" is different thing than dojo grid.

Comment 5

[alexander :surkov (:asurkov)]

Attachment: patch

Comment 6

[David Bolter [:davidb] (NeedInfo me for attention)]

Comment on attachment 400238 [details] [diff] [review]
patch

r=me thanks, with nits:

> #ifdef DEBUG
>-            nsRoleMapEntry *tableRoleMapEntry =
>-              nsAccUtils::GetRoleMapEntry(tableNode);
>-            NS_ASSERTION(tableRoleMapEntry &&
>-                         !nsCRT::strcmp(tableRoleMapEntry->roleString, "presentation"),
>-                         "No accessible for parent table and it didn't have role of presentation");
>+          nsRoleMapEntry *tableRoleMapEntry =
>+            nsAccUtils::GetRoleMapEntry(tableNode);
>+          NS_ASSERTION(tableRoleMapEntry &&
>+                       !nsCRT::strcmp(tableRoleMapEntry->roleString, "presentation"),
>+                       "No accessible for parent table and it didn't have role of presentation");
> #endif

I'd prefer to leave this minor formatting change out (since it messes up the hg annotations).


>+          if (!roleMapEntry && !content->IsFocusable()) {
>+            // Table-related descendants of presentation table are also
>+            // presentation if they aren't focusable and have not explict ARIA

Nit: "have no explicit"

Comment 7

[alexander :surkov (:asurkov)]

(In reply to comment #6)

> I'd prefer to leave this minor formatting change out (since it messes up the hg
> annotations).

they have different indenting.

Comment 8

[Marco Zehe (:MarcoZ)]

Comment on attachment 400238 [details] [diff] [review]
patch

r=me thanks! No further nits than what David said.

Comment 9

[David Bolter [:davidb] (NeedInfo me for attention)]

(In reply to comment #7)
> (In reply to comment #6)
> 
> > I'd prefer to leave this minor formatting change out (since it messes up the hg
> > annotations).
> 
> they have different indenting.

If hunting down the blame/credit (annotation) could go deeper than 1 level I wouldn't care :)  But I rely on it a lot.

Comment 10

[alexander :surkov (:asurkov)]

(In reply to comment #9)

> If hunting down the blame/credit (annotation) could go deeper than 1 level I
> wouldn't care :)  But I rely on it a lot.

usually hunting down is always deeper than 1 :) but it sounds it didn't catch your idea.

Comment 11

[alexander :surkov (:asurkov)]

landed on 1.9.3 - http://hg.mozilla.org/mozilla-central/rev/00bc9fd43693

Comment 12

[David Bolter [:davidb] (NeedInfo me for attention)]

Alexander, Marco, do we want to try and squeeze this into 1.9.2?

Comment 13

[alexander :surkov (:asurkov)]

We could. However we weren't in time to introduce all table related changes in 1.9.2 and continued the work in 1.9.3 (including IA2 table2 interface support which is the major landmark). We can think we just are faced to the fact: Firefox 3.6 is not entirely accessible from point of view of table work.

However if we'll land this then we could improve table navigation methods for dojo grids I think. This might be worth.

Comment 14

[Marco Zehe (:MarcoZ)]

Comment on attachment 400238 [details] [diff] [review]
patch

This has been baking on m-c for a while without causing regressions. Needed for completion of a11y support for tables in 3.6. See https://wiki.mozilla.org/Accessibility/Remaining_Mozilla-1.9.2_Nominations for more info.

Comment 15

[Mike Beltzner [:beltzner, not reading bugmail]]

Comment on attachment 400238 [details] [diff] [review]
patch

a192=beltzner

Comment 16

[alexander :surkov (:asurkov)]

landed on 1.9.2 - http://hg.mozilla.org/releases/mozilla-1.9.2/rev/dc79d372f816

Comment 17

[Marco Zehe (:MarcoZ)]

Verified fixed in Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.2b2pre) Gecko/20091029 Namoroka/3.6b2pre (.NET CLR 3.5.30729)