Apache OpenOffice (AOO) Bugzilla – Issue 81824
[a11y] Orca is too chatty when navigating by paragraph in OOo Writer
Last modified: 2013-08-07 14:43:11 UTC
See Orca bug #435201 which has been reopen and is blocked by this issue. http://bugzilla.gnome.org/show_bug.cgi?id=435201 This is with OpenOffice 2.3.0 RC1 (ubuntu2) on latest Gutsy. A simple OOo file to use -- spanish.odt -- can be found at: http://bugzilla.gnome.org/attachment.cgi?id=95966&action=view Steps to reproduce: 1/ Start Orca 2/ Start oowriter spanish.odt 3/ Type Control-home to get to the top of the document. 4/ Type Control-down to get to the second paragraph. This first line is spoken correctly. 5/ Type Control-down to get to the first blank paragraph. 6/ Type Control-down to get to the next paragraph. The first line is spoken three times. See our (the Orca team's) analysis at: http://bugzilla.gnome.org/show_bug.cgi?id=435201#c23 and http://bugzilla.gnome.org/show_bug.cgi?id=435201#c24
Will asked me to ask you (the OOo a11y team) what your intended meaning of a detail=-1 on a caret-moved event is? Thanks.
The other thing I forgot to mention is that when running the steps to recreate this problem, the following is output to the terminal window in which I started oowriter: ** (soffice:17642): WARNING **: Exception in getText() ** (soffice:17642): WARNING **: Exception in getText() ** (soffice:17642): WARNING **: Exception in getText() ** (soffice:17642): WARNING **: Exception in getText()
Two main comments to add here to hopefully help progress on this bug: 1) I'm assuming a object:text-caret-moved event with a detail=(-1,0) means that the caret is no longer positioned anywhere in the text. 2) I think we might be getting a spurious focus event from OOo when we shouldn't be, and that seems to be what is causing the excessive chattiness. Please see also http://bugzilla.gnome.org/show_bug.cgi?id=435201#c30 for an example sequence of events delivered by OOo when navigating by paragraph. Thanks!
Yes, a caret position of '-1' means the caret is no longer in this particular paragraph object. The exception warning should not hurt, but I will have a look at it. On the spurious focus events: what you are observing (at least according to the log you referenced) are state change events. Those are by design dispatched synchronously, in contrast to the global focus events which are only sent when idle. So it seems the empty paragraph temporarily gains the keyboard focus and I doubt this behavior can easily be changed in writer. Can't you just ignore state change events for "focused" and rely on the global focus events ?
>Yes, a caret position of '-1' means the caret is no longer in this >particular paragraph object. Thanks! > The exception warning should not hurt, but I will have a look at it. It's scary, though, to users who don't expect these kinds of things. :-) > So it seems the empty paragraph temporarily gains the keyboard focus > and I doubt this behavior can easily be changed in writer. Bummer. :-( > Can't you just ignore state change events for "focused" and rely on the > global focus events ? I see. We might be able to do this, though we have various bugs in various toolkits that we've needed to work around and it requires us to handle both the focus: and state-changed:focused events. At this point, however, we've been able to work around the spurious event. If you have a way to prevent it from being issued, however, it would be great. Thanks for your work on this!
As described: the 3rd text paragraph "Now old lady" is reapated 3 times, the second "NOBODY expects" is repeated 2 times.
STARTED, OOo 3.1
Evaluation showed that in case of traveling to another paragraph using Ctrl-up/down SwAccessibleParagraph::_InvalidateCursorPos() is called 6 times (3 times focus lost, 3 times focus gained) whereas when just traveling with simple Cursor-up/down it's only called twice correctly. ->od as discussed
fixed in cws sw31a11y01 - changed file: /sw/source/ui/wrtsh/wrtsh4.cxx, rev. 265904 defect cause: unnecessary cursor movements to move the cursor to the previous/next paragraph.
OD->ES: Please verify in internal installation
Verified in CWS sw31a11y01
Ok in OOO310m5