391482 - When ARIA state of invalid is set, screen readers speak the wrong thing

Status: RESOLVED FIXED

(Core :: Disability Access APIs, defect)

Description

[Pete Brunet]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9a7pre) Gecko/2007070504 Minefield/3.0a7pre
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9a7pre) Gecko/2007070504 Minefield/3.0a7pre

Using the test URL tab down to the Occupation field.  When it has no data it has an ARIA state of INVALID, but the MSAA state of STATE_SYSTEM_ALERT_HIGH is not exposed.  (There is no problem on FF2, 2.0.0.6)

Reproducible: Always

Steps to Reproduce:
1.See details.
2.
3.
Actual Results:  
STATE_SYSTEM_ALERT_HIGH is not set

Expected Results:  
STATE_SYSTEM_ALERT_HIGH is set

Comment 1

[Pete Brunet]

Changed from Firefox to Core as suggested by Aaron Leventhal.

Comment 2

[Aaron Leventhal]

Attachment: Testcase -- works in FF2 and FF3

Comment 3

[Aaron Leventhal]

I looked at the Dojo example in DOM Inspector, and invalid is not set. Consistent with this, the example does not work in Firefox 2.

Comment 4

[Pete Brunet]

Aaron, Apparently the nightly is not up to date yet with the code we committed to set the ARIA state of invalid.  I see a failure on my local dojo build, but it's not what I originally reported.  ALERT_HIGH is reported by Inspect on both FF2 and FF3.  However there is a problem with Window-Eyes and FF3.  On FF2 W-E announces "invalid" but not on FF3.  The situation JAWS is different.  I hear ALERT on FF3.  I think this is related to the popup.  I don't hear ALERT or INVALID with JAWS on FF2.

Comment 5

[Aaron Leventhal]

Hi Pete, can you attach a test case that shows the problem?

I've renamed the bug. Does it sound like an accurate statement of the problem?

Comment 6

[Pete Brunet]

Aaron, The minimal test case you provided is OK.  

Unfortunately I am getting inconsistent results from day to day.  Today I notice that W-E on FF2 and FF3 is not announcing "invalid" with browse mode off but it does announce it with browse mode on.  That's with my local build.  The nightly link is still not reflecting the latest dojo code.

I am not concerned about JAWS right now as I think they have work to do in general.  (Is that correct?)

I think it would be OK to close this bug since I hear "invalid" in at least browse mode on both FF2 and FF3.

Comment 7

[Aaron Leventhal]

Tim, can you file a bug for both JAWS and Window-Eyes, for the minimal testcase attached to this bug?

The textfield should be reported as invalid (not alert) for all modes, browse/virtual or forms mode.

When done please close this bug as it's not a bug in the Firefox code.

Comment 8

[Aaron Leventhal]

Actually a bug is already filed for JAWS, and we may only need to file one for Windoiw-Eyes. In fact, bug 391592 might fix this since we weren't exposing the IA2 states correctly. So let's wait until that ones fixed and see what happens.

Comment 9

[Aaron Leventhal]

Pete, do you mind trying with Window-Eyes and JAWS again with a recent build of Firefox now that bug 391592 is fixed?

Comment 10

[Pete Brunet]

Aaron, Using W-E 6.1 (April 11, 2007) with 3.0a8pre and http://archive.dojotoolkit.org/nightly/dojotoolkit/dijit/tests/form/test_validate.html I see the ALERT_HIGH state in Inspect and I heard "invalid" in browse mode but I don't hear it when browse mode is off.

Comment 11

[Aaron Leventhal]

I'll file a bug in the Window-Eyes database.