As an experiment, I put a "return utterances" line at the beginning of clumpUtterances and things *seemed* to work OK, though I didn't run the regression tests.

(In reply to comment #1)
> As an experiment, I put a "return utterances" line at the beginning of
> clumpUtterances and things *seemed* to work OK, though I didn't run the
> regression tests.

Can you spell "NOPE!"?  :-)

Needs some more investigation.  Here's an example regression:

Test 5 of 13 FAILED: /export/home/orca/orca/test/keystrokes/firefox/bug_512303.p
y:Line Down
DIFFERENCES FOUND:
  BRAILLE LINE:  '4 5 6'
       VISIBLE:  '4 5 6', cursor=1
- SPEECH OUTPUT: '4 5 6'
?                  ----

+ SPEECH OUTPUT: '4'
+ SPEECH OUTPUT: '5'
+ SPEECH OUTPUT: '6'

*** Bug 586078 has been marked as a duplicate of this bug. ***

See also http://bugzilla.gnome.org/show_bug.cgi?id=586078#c2 for how clumpUtterances is now buggy since it assumes utterances are composed of strings only.

Created attachment 136831 [details] [review]
Patch to avoid issues with clumpUtterances

This patch hopefully avoids the issues with clumpUtterances for now until we decide what to do with it.

This patch is perfect!
Orca+Tab and Orca+Shift+tab combinations work again, Orca speech correct the next or prewious form fields.

Attila

Thanks for testing, Attila!