This is a more general problem. Orca just speaks 'scroll pane' when an internal frame gets keyboard focus.

Correction: Orca speaks 'scroll pane' only if the internal frame is empty. Otherwise, Orca speaks the component that has focus in the internal frame that has focus.

Question: What is the correct behavior when a different internal frame first gets focus?

Control-F6 is used to navigate among the internal frames in a window. Tab is used to navigate among the components in the internal frame that has focus.

Right now, Orca just speaks the component that got focus in the internal frame that got focus. I propose that when a different internal frame first gets focus, Orca should speak the name of the internal frame followed by the name of the component in the internal frame that got focus. Otherwise, the user has no indication that she is navigating inside a different internal frame.

You probably should take a look at the panel and frame examples from the gtk-demo application for an example of how this is intended to work.

Mike,

Would you be more explicit about the names of the demos I should look at. I could find anything like internal frames in the gtk-demo examples.

Thanks

Created attachment 91712 [details] [review]
potential patch

Assuming proposal in Comment #2 is acceptable, the attached patch implements this behavior.

The typical way Orca handles the presentation of text associated with containers is via speechgenerator.py:getSpeechContext and braillegenerator.py:getBrailleContext.  For some reason, it seems like these might either not be getting called or they might be skipping JInternalFrames for some reason.  Can you take a look in this area of the code and see what's going on?

Created attachment 91875 [details] [review]
potential patch

It appears that line 1658 in the speechgenerator.getSpeechContext method is too restrictive. The default.Script.getDisplayedText method may use the parent name if the parent text object is None. 

Orca speaks the JInternalFrame context correctly with this patch.

(In reply to comment #7)
> Created an attachment (id=91875) [edit]
> potential patch
> 
> It appears that line 1658 in the speechgenerator.getSpeechContext method is too
> restrictive. The default.Script.getDisplayedText method may use the parent name
> if the parent text object is None. 
> 
> Orca speaks the JInternalFrame context correctly with this patch.
> 

Definitely an interesting patch.  Is the root of the problem that JInternalFrame does not implement Accessible Text (seems like it should), and that it sets the Accessible name to the text being displayed?

See also http://bugzilla.gnome.org/show_bug.cgi?id=458142#c3 of bug 458142 for comments on this patch.  Thanks!

*** This bug has been marked as a duplicate of 458142 ***