542726 - [SeaMonkey 2.1] mochitest-a11y: test_name_nsRootAcc.xul fails

Status: VERIFIED INVALID

(Core :: Disability Access APIs, defect)

Description

[alexander :surkov (:asurkov)]

The problem is when we get focus event the title bar is not updated yet. It doesn't happen with Firefox because Firefox and Seamoneky tabbrowsers handle "select" event differently (http://mxr.mozilla.org/mozilla-central/source/browser/base/content/tabbrowser.xml#69 and http://mxr.mozilla.org/comm-central/source/suite/browser/tabbrowser.xml#2386. So that Seamonkey handles "select" after "focus".

Comment 1

[alexander :surkov (:asurkov)]

Moreover nsIInterfaceRequestor is not defined - http://mxr.mozilla.org/comm-central/source/suite/browser/tabbrowser.xml#790. I think it's assumed it's defined somewhere outside of binding. But it's not good in general and I can see Components.interfaces.nsIInterfaceRequestor usage in another place of tabbrowser.xml so I think this could be fixed in tabbrowser.xml.

Also I get two warnings: "Warning: Empty string passed to getElementById()."

But since neither file name nor the line number are provided then I have no idea what could be wrong. Possibly it doesn't affect on the test.

Comment 2

[alexander :surkov (:asurkov)]

Neil, could the "select" event be handled the same way like in firefox's tabbrowser?

Comment 3

[alexander :surkov (:asurkov)]

Neil, could tabbrowser.xml be shared between Firefox and Seamonkey? for example be moved to toolkit. I realize it might be trick to get rid browser internal dependency from tabbrowser binding but it should be worth.

Comment 4

[[e-mail]]

(In reply to comment #1)
> Moreover nsIInterfaceRequestor is not defined -
> http://mxr.mozilla.org/comm-central/source/suite/browser/tabbrowser.xml#790. I
> think it's assumed it's defined somewhere outside of binding. But it's not good
> in general and I can see Components.interfaces.nsIInterfaceRequestor usage in
> another place of tabbrowser.xml so I think this could be fixed in
> tabbrowser.xml.
Yes, that's a bug. I'll r+sr a patch.

> Also I get two warnings: "Warning: Empty string passed to getElementById()."
> 
> But since neither file name nor the line number are provided then I have no
> idea what could be wrong. Possibly it doesn't affect on the test.
You really need a debugger to take a stack trace for this one. (And the caller is likely to be JS, so you'll have to take a DumpJSStack() too.)

Comment 5

[[e-mail]]

(In reply to comment #0)
> The problem is when we get focus event the title bar is not updated yet. It
> doesn't happen with Firefox because Firefox and Seamoneky tabbrowsers handle
> "select" event differently
> (http://mxr.mozilla.org/mozilla-central/source/browser/base/content/tabbrowser.xml#69
> and
> http://mxr.mozilla.org/comm-central/source/suite/browser/tabbrowser.xml#2386.
> So that Seamonkey handles "select" after "focus".
Not sure what you mean there, the select event bubbles up from the tabbox to the tabbrowser so it should fire almost exactly at the same time.

(In reply to comment #3)
> Neil, could tabbrowser.xml be shared between Firefox and Seamonkey? for example
> be moved to toolkit. I realize it might be trick to get rid browser internal
> dependency from tabbrowser binding but it should be worth.
Unfortunately they are too different, which is why it's not in toolkit.

Comment 6

[alexander :surkov (:asurkov)]

(In reply to comment #5)
> (In reply to comment #0)
> > The problem is when we get focus event the title bar is not updated yet. It
> > doesn't happen with Firefox because Firefox and Seamoneky tabbrowsers handle
> > "select" event differently
> > (http://mxr.mozilla.org/mozilla-central/source/browser/base/content/tabbrowser.xml#69
> > and
> > http://mxr.mozilla.org/comm-central/source/suite/browser/tabbrowser.xml#2386.
> > So that Seamonkey handles "select" after "focus".
> Not sure what you mean there, the select event bubbles up from the tabbox to
> the tabbrowser so it should fire almost exactly at the same time.

Yes, but it sounds this "almost" changes the things. We don't handle "select" for tabbox in a11y so I think we handle focus event after tabbox handles "select" and before "tabbrowser" handles it. At least this sounds verisimilar and I don't have other explanation. I could check to be ensure.

Comment 7

[alexander :surkov (:asurkov)]

(In reply to comment #4)

> You really need a debugger to take a stack trace for this one. (And the caller
> is likely to be JS, so you'll have to take a DumpJSStack() too.)

Can you please give details how I should use DumpJSStack()?

Comment 8

[[e-mail]]

Oh, I think the issue might be that we don't set the title on the XUL document.

(In reply to comment #7)
> Can you please give details how I should use DumpJSStack()?
In Windows, you need to run with -console, then with WinDbg use .call xpc3250!DumpJSStack() and then g (or press F5), while with Visual Studio add {,,xpc3250}DumpJSStack() to the Watch window (remember to remove it again).
On Linux, assuming you've already loaded the symbols, just call DumpJSStack()

Comment 9

[Serge Gautherie (:sgautherie)]

http://tinderbox.mozilla.org/showlog.cgi?log=SeaMonkey/1265177678.1265179466.17592.gz&fulltext=1#err35
Linux comm-central-trunk debug test mochitest-other on 2010/02/02 22:14:38
{
2616 ERROR TEST-UNEXPECTED-FAIL | chrome://mochikit/content/a11y/accessible/test_name_nsRootAcc.xul | Test timed out.

mochitest-a11y 22376/254271/2
}

Ftr, this test reports "a lot" of warnings and assertions, but I'm not sure if any are relevant to this issue.

Comment 10

[Serge Gautherie (:sgautherie)]

Bug 525175 disabled this test on Linux.

Comment 11

[Justin Wood (:Callek)]

and disabled a long while ago...

Comment 12

[Serge Gautherie (:sgautherie)]

(In reply to comment #4)
> (In reply to comment #1)
> > Moreover nsIInterfaceRequestor is not defined -
> Yes, that's a bug. I'll r+sr a patch.

I assume this was bug 591550.

(In reply to comment #11)
> and disabled a long while ago...

Fully disabled a month ago only, by bug 586818.

V.Invalid, at least until the test is re-enabled.

Comment 13

[Serge Gautherie (:sgautherie)]

Ftr, bug 719754 rewrote this test.