GNOME Bugzilla – Bug 608150
Firefox 3.6's changes to the hierarchy of tables needs to be handled by Orca
Last modified: 2010-08-01 23:09:49 UTC
Dear developers, After my Lucid system upgraded with Firefox 3.6, Orca does'nt spokening some form fields with webpages, for example bugzilla.gnome.org. When I report a new bug, choosing Orca component and press Capslock+Tab key, Orca does'nt spokening the version list view. Another example: When I try add an attachment with a bugreport, Orca does'nt spokening the file edit box. Prewyous Firefox version don't do this, works very well with Orca. I sending you the debug.out file. For example, when I setting attachment informations, happen following before I go to the browse... button with tab key: Not copy full informations, only the generate speech result and traceback part, possible help you: generate speech for app.name='Firefox' name='File' role='entry' state='enabled focusable focused horizontal opaque sensitive showing single line visible' relations='' (args={'useDefaultFormatting': False, 'formatType': 'unfocused', 'role': ROLE_ENTRY, 'mode': 'speech', 'recursing': True}) using 'newAncestors + newRowHeader + newColumnHeader + newRadioButtonGroup + labelOrName + readOnly + textRole + currentLineText + allTextSelection + (mnemonic and (pause + mnemonic + lineBreak) or []) + newNodeLevel + unselectedCell + (tutorial and (pause + tutorial) or [])' newAncestors=[] newRowHeader=[] newColumnHeader=[] newRadioButtonGroup=[] Traceback (most recent call last):
+ Trace 220263
globalsDict[arg] = self._methodsDict[arg](obj, **args)
self, obj, **args))
result.extend(self._generateLabel(obj, **args))
label = self._script.guessTheLabel(obj)
guess = self.guessLabelFromTable(obj)
self.getNextCellInfo(containingCell, "left")
table = cell.parent.queryTable()
raise NotImplementedError NotImplementedError
generate speech results: Attila Attila
I changing this bug status to critical, because impossible using right with Firefox 3.6 web browser if this bug is present. Attila
I can also repeat this type problem. I described it in more detail in a different bug before I knew about this one.
After upgrade to firefox 3.6 I tried to access www.google.com. I could not leave the form using the arrows. Every time I reached the end of the form I went back to the top of it.
(In reply to comment #3) > After upgrade to firefox 3.6 I tried to access www.google.com. > I could not leave the form using the arrows. Every time I reached the end of > the form I went back to the top of it. Sounds like bug 608149. I'm working on it.
Guys, I could use some testing help here. Using Orca from git master, and having unchecked the new option related to grabbing focus, are the issues reported in this bug still present? (As a reminder to potential testers, you only want to uncheck the grab focus checkbox if you are using Firefox 3.6 from 30th September or later.)
I've an internal application in which I can reproduce the problem using firefox 3.6 and last orca from git. It's my time sheet, I need to fill every day. I can not give a link to my application but I will attach a html that can be used to reproduce the problem. Unfortunately, the content is in Portuguese, which can cause some difficulty in debugging.
Created attachment 153308 [details] html file in which I can notnavigate pressing tab key orca does not announce the content and the label of the fields when navigating with the tab key
Taking this one. (In reply to comment #1) > I changing this bug status to critical, because impossible using right with > Firefox 3.6 web browser if this bug is present. A 'critical' bug is defined as 'Crashes, causes loss of data, or is a severe memory leak'. That is not the case here. A 'major' bug is defined as 'Major loss of functionality - menu item broken, data output extremely incorrect, or otherwise difficult/useless to use.' I think we can all agree on the 'otherwise difficult/useless to use'. Therefore I'm lowering the severity of this bug to major.
Created attachment 153803 [details] [review] part -- but not all -- of the fix One issue is that table rows can now appear as list items in the hierarchy. getNextCellInfo() was assuming the parent of the table cell would be a table. This fix merely stops the traceback by correctly locating the table. It does *not* cause Orca to say the right thing. For example, in Firefox 3.6, Orca is now always reporting "Del." as the field label in the sample timesheet. To handle this situation, additional smarts will need to be added to Orca's label guessing functionality. I'll tackle that next. (This patch has already been committed to master.)
Created attachment 153804 [details] [review] rest of the fix It turns out that the label guessing functionality is fine (yea!!); the problem is another instance of needing to be sure we have the table and not the new list item. This change causes us to do the right thing with the timesheet example both in Firefox 3.5 and Firefox 3.6. Because of this, and because it's a conservative change, I've gone ahead and committed it to master.
You'll notice that I re-summarized this bug to reflect a very specific problem related to tables. The traceback in Attila's opening report matches what I was seeing in Jose's test case. Because I have fixed that specific problem, I'm closing this bug as FIXED. If there are other problems with guessing labels which were introduced by Firefox 3.6, a new bug should be opened. This bug should *not* be re-opened unless the issue is table-related breakage introduced by FF 3.6 or the fix I've provided here have introduced a regression. Thanks!