GNOME Bugzilla – Bug 439286
Gecko.py's find{Next,Previous}Object fails when object is document_frame
Last modified: 2008-07-22 19:27:24 UTC
In the test case to be attached, move focus to any of the lines of text that are NOT immediate children of the document frame, i.e. * Some text * More text * Last of the text Having done so, try using H to move to the next heading. Focus will be moved to the top of the document first, then the heading at the top of the document will be found. A similar problem can be seen when using T to move to the next table. (Try it from the "More text" line) Having done so, instead try using Shift+H or Shift+T to move to the previous heading or table. Orca will not find any headings or tables.
Created attachment 88361 [details] aforementioned test case
Created attachment 88362 [details] [review] first pass at this I *think* I've solved the problem. That said, given that find{Next,Previous}Object() are such significant methods -- complete with the possibility of endless looping.... Will please review this when you have a chance, and are not in a hurry, and are fully equipped to spot the stuff I've neglected to overlook. ;-) ;-) ;-) Thanks much!!
> Will please review this when you have a chance, and are not in a hurry, and are > fully equipped to spot the stuff I've neglected to overlook. ;-) ;-) ;-) Joanie and I talked about this on the phone. I kind of grok what is going on, though it's obvious Joanie understands Gecko.py more than I do now. :-) The main area of impact I see regarding this change is that it touches the core of any structural navigation. So, I guess the main thing to do here would be to do some thorough testing with all aspects of structural navigation. If that continues to work, then I say commit. :-)
> main area of impact I see regarding this change is that it touches the core of > any structural navigation. So, I guess the main thing to do here would be to > do some thorough testing with all aspects of structural navigation. If that > continues to work, then I say commit. :-) While I've done quite a bit of testing, I haven't done what I would consider to be "thorough" testing. I will do that. Mike, can I also ask you to pound on this -- and *really* pound on it? :-) Please pay special attention to all of the previousWhatever structural navigation commands, including starting from the very end of the page and also when you know there are no more previousWhatevers. If anything is going to break, that's likely where it's going to break. Thanks!
Created attachment 88456 [details] [review] minor change One of the methods added in this patch, getLastObject(), is also needed in the RFE for wrapping structural navigation so I checked that method in. This version of the patch is the same as the other, only without getLastObject() so you won't get the hunk failed/rejected error.
I've been testing this patch on quite a variety of sites over the weekend. It does not seem to be failing in the ways Joanie told me I might see. I quite possibly have missed something but feel that I've given this patch pretty detailed testing.
This seems ready for closing. Joanie?
Sounds good. Just committed the patch. Thanks!