Hi Joanie,
Thanks for these technical precisions, helpful to report bugs.
I have, at the end of this mail threads, a global question: let's suppose we can fix all the accessibility tree (probably an ambitious work, but well). To implement non-regression tests, I imagine two kind of tests. - Testing the presence or not of labels, eg in Glade, with test of positive talse, and maybe a file to gather the labels in order to determine how they are relevant. - Testing the presence of link between the widgets in the accessibility tree. And if present, as for labels, creating a kind of ifile to check they are relevant.
Is it automatisable enough you think? Or ingenious?
It would be a good balance between simple tests and scenarios tests if it is technically acceptable.
Regards
Le 07/08/2017 à 20:14, Joanmarie Diggs
a écrit :
So here's my conclusion from attempting to hack around it: 1. Orca says the frame name because the frame gets activated. At this point, Orca doesn't know about the dialog. And diving down the accessibility tree looking for a dialog just in case one is there has the potential to be non-performant and lead to great delays in responsiveness in frames with lots and lots of descendants. If the dialog were an immediate descendant of the frame, this would suck less. But it's inside a panel which is inside another panel. We don't want to do such a just-in-case dive in Orca. Doubly so if the only benefit is to work around a bit of chattiness. 2. I can filter out the presentation of the dialog because that's presented as an ancestor of the newly-focused widget (different accessibility event). However, that means Orca presents the frame and then the focused widget and ancestors, but never tells you you're in a dialog. Knowing that you're in a dialog can be useful because it tells you things about keystrokes you can expect to work (e.g. Escape to close the dialog, the possible presence of a default button triggered by pressing Enter, etc.). We don't want to prevent the user from knowing about these things. Doubly so if the only benefit is to work around a bit of chattiness. Therefore, I'm not going to hack around this one in Orca. And hopefully the above information combined with your ability to use Accerciser will give you what you need to file this bug. I'll leave the broken ancestry / index in parent of -1 on my todo list for now. --joanie On 08/07/2017 01:22 PM, Joanmarie Diggs wrote:The dialog is contained in a frame with the same name. I'm trying to hack around that now both for the chattiness issue and because of your report that Orca is not presenting the focused object in the Print dialog/frame. Looks like there's also some broken hierarchy stuff going on with an index in parent of -1. I'll add reporting these issues to my todo list, but this looks like a bit of a mess on multiple fronts in LibreOffice and something which my hacking around may lead to more regressions than improvements. --joanie On 08/07/2017 12:58 PM, Alex ARNAUD wrote:Le 07/08/2017 à 18:43, Joanmarie Diggs a écrit :Hi Alex. My guess is that the dialog is contained inside a frame and both are new. I can see about filtering one of those out, but doing so would be a hack and would only work if the names are exactly the same. Which they may be in this particular case, but can we always count on that?So if I understand well dialog shouldn't contain frame as it child ?It would be nice if LibreOffice could prune some of that noise out of their own accessibility tree.I don't know exactly what to report upstream here. Could you report this bug or explain me? This issue is old, I can reproduce it on LibreOffice 4.2. Best regards._______________________________________________ orca-list mailing list orca-list gnome org https://mail.gnome.org/mailman/listinfo/orca-list Orca wiki: https://wiki.gnome.org/Projects/Orca Orca documentation: https://help.gnome.org/users/orca/stable/ GNOME Universal Access guide: https://help.gnome.org/users/gnome-help/stable/a11y.html Log bugs and feature requests at http://bugzilla.gnome.org_______________________________________________ orca-list mailing list orca-list gnome org https://mail.gnome.org/mailman/listinfo/orca-list Orca wiki: https://wiki.gnome.org/Projects/Orca Orca documentation: https://help.gnome.org/users/orca/stable/ GNOME Universal Access guide: https://help.gnome.org/users/gnome-help/stable/a11y.html Log bugs and feature requests at http://bugzilla.gnome.org --
|