There are some good reasons to think we should get rid of this dependency in accessibility/atk, such as the fact that we need to properly expose wrapped lines in WebKit2 (see https://bugs.webkit.org/show_bug.cgi?id=73433#c1) or that similar results to what we get through pango/gail combo could be achieved using internal utilities inside WebKit (e.g. TextBreakIterator).
However, this is probably not going to be a one-patch-fix, so here is the metabug to track them all.
Changing the nature of this metabug to convert it into a normal bug (hence renaming it), so we can attach to it the final patch that will remove the pango/gail code from WebKitAccessibleInterfaceText.cpp, once all the blocking bugs have been solved.
Created attachment 207192 [details]
...and here comes the final patch!
Comment on attachment 207192 [details]
View in context: https://bugs.webkit.org/attachment.cgi?id=207192&action=review
> + return resultText.release();
I think if you always release the resultText then perhaps just a raw pointer works.
(In reply to comment #3)
> (From update of attachment 207192 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=207192&action=review
> > Source/WebCore/accessibility/atk/WebKitAccessibleInterfaceText.cpp:1075
> > + return resultText.release();
> I think if you always release the resultText then perhaps just a
> raw pointer works.
Yeah, you're probably right. I'll change that before landing, once than the patches for the blocking bugs have been landed.
Committed r155520: <http://trac.webkit.org/changeset/155520>