GNOME Bugzilla – Bug 470332
Can no longer arrow to push buttons in FF3
Last modified: 2008-07-22 19:28:27 UTC
Step to reproduce: 1. Try to move to a push button using the arrow keys. Expected Results: Orca would let you arrow to the push button. Orca would speak lines that contained push buttons. Actual Results: Orca skips right over push buttons. Notes: This might be a side effect of some other changes to FF3. I recall at one point there was some discussion about the possibility of all objects implementing text and hypertext because there is the potential for an object to start out devoid of text/hypertext and but to gain it dynamically. Our problem seems to be that push buttons are implementing text, but have a characterCount of 0. Patch to follow.
Created attachment 94342 [details] [review] proposed patch Just turned a bunch checks on obj.text to checks on obj.text and obj.text.characterCount. Seems to solve the problem. Will please review.
(In reply to comment #1) > Created an attachment (id=94342) [edit] > proposed patch > > Just turned a bunch checks on obj.text to checks on obj.text and > obj.text.characterCount. Seems to solve the problem. Will please review. > Looks good. Thanks!
I have been running with this patch now, and am noticing that I can now arrow to the buttons, but they do not yet show up in braille. For example, on www.google.com, the Google Search and I'm Feeling Lucky buttons are both spoken, but not brailled.
Created attachment 94378 [details] [review] D'oh! Also do the check in updateBraille() Sleep, I don't need no stinking sleep. Then again, apparently I do.... :-) This version also does the check in updateBraille() and should solve the issue Marco addressed. Thanks for testing this Marco! Mind giving this version a spin?
Given that the new version is merely adds a couple of additional checks for actual characters and also seems to work <grin>, and given that we have a release to roll out, I went ahead and committed the revised version both to trunk and the gnome-2-20 branch. So, Marco, in order to test it now, you should reverse the original patch and *then* do the svn update. Thanks again!!
Looks good. This also fixes the speaking/brailling of buttons when using form field navigation.
Great thanks! Moving to pending.
Moving this out of pending. It seems to have a nasty side effect with lists. Still investigating....
Created attachment 94471 [details] [review] one more text.characterCount check The bad list juju is a Mozilla bug (which we're now tracking via bug #470853). Among other things, we're getting a bogus characterCount == 0 for lists. Until the Mozilla bug is fixed, this defensive patch should hopefully do the trick. I won't be so bold as to jinx us by saying it's just another harmless check.... ;-)
Okay, patch checked in to both trunk and the gnome-2-20 branch. Marco, Mike: Please test. Will, hackishness aside, anything I'm missing?
Yech to the Mozilla bug, but hooray for you for the defensive maneuver. My only question was if we'd end up with a non-None text but a None obj.parent.text. In looking at getUnicodeText, however, this doesn't seem like it would ever be the case. So, looks cool to me. Thanks!
This seems to work nicely.
Thanks! Closing as FIXED.