843657 - Verify that the Plugin Hang UI is working with screen readers

Status: RESOLVED FIXED

(Core Graveyard :: Plug-ins, defect, P2)

Description

[(No longer employed by Mozilla) Aaron Klotz]

A user reported on Twitter that the Plugin Hang UI was not being recognized by his screen reader (JAWS 13). However, he also indicated that he was experiencing hangs and focus problems with Firefox when running JAWS.

I downloaded the JAWS 14 evaluation and tried it on Nightly. As soon as flash loaded, Firefox hung (probably bug 781971).

Once bug 781971 is fixed, we should review how the Plugin Hang UI behaves with screen readers.

Comment 1

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

Thanks for filing.

cc+ MarcoZ.

We should test the plugin hang UI with NVDA which is not susceptible to bug 781971 to check if this is a plugin hang UI isolated problem.

Comment 2

[Marco Zehe (:MarcoZ)]

Is there a way to provoke a plugin hang somehow to test if the UI works?

Comment 3

[Benjamin Smedberg]

You can use sysinternal process explorer to suspend plugin-container, or you can use the testplugin .hang() method.

Comment 4

[(not currently active) Ted Mielczarek]

bug 805591 comment 40 has directions on how to get and use the test plugin.

Comment 5

[(No longer employed by Mozilla) Aaron Klotz]

(In reply to David Bolter [:davidb] from comment #1)
> We should test the plugin hang UI with NVDA which is not susceptible to bug
> 781971 to check if this is a plugin hang UI isolated problem.
NVDA worked fine for me.

Comment 6

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

(In reply to Aaron Klotz [:aklotz] from comment #5)
> (In reply to David Bolter [:davidb] from comment #1)
> > We should test the plugin hang UI with NVDA which is not susceptible to bug
> > 781971 to check if this is a plugin hang UI isolated problem.
> NVDA worked fine for me.

Thanks. By the way is this bug still presenting with Jaws?

Marco can you own/drive this one?

Comment 7

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

I'm also getting hangy behaviour with Flash and Jaws 14 eval.

Comment 8

[Marco Zehe (:MarcoZ)]

OK, I've tried the first suggestion:

(In reply to Benjamin Smedberg  [:bsmedberg] from comment #3)
> You can use sysinternal process explorer to suspend plugin-container,

and found that Process Explorer is not accessible. The list items don't read with NVDA or JAWS.

> or you can use the testplugin .hang() method.

I need more info on that, what I need etc. to be able to execute that method.

Comment 9

[(not currently active) Ted Mielczarek]

Marco: if you have a locally-built Firefox and you run it from the objdir it should have the test plugin available. You can then use this test page to hang it:
http://people.mozilla.com/~tmielczarek/testplugincrash.html

If you don't have a local build, bug 805591 comment 40 has directions on how to get and use the test plugin.

Comment 10

[Marco Zehe (:MarcoZ)]

The plugin hang UI is, in principle, accessible. That means the mark-up and widgets used are correct.

HOWEVER, even NVDA becomes very unresponsive once the plugin hang UI comes up. The only thing that did update was my braille display. Speech was no longer able to give me any feedback. The initial dialog text was spoken by NVDA, although that may have just been a favorable timing-related coincidence.

In other words: it is accessible, but the plugin UI itself and the screen reader become very unresponsive, and in the case of NVDA, allow for only a braille display to update once tabbing through the UI. Since many users won't have a braille display to see that there is any reaction at all, to them it will appear as if the screen reader and browser have totally frozen.

Comment 11

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

That seems to imply the hang is between the (probably in-process) screen reader and the plugin. I reached out to Freedom Scientific via email.

Comment 12

[Marco Zehe (:MarcoZ)]

Last time Flash hang for me, which is ages ago, I saw this working. Closing.