Created attachment 526222 [details] [diff] [review] Patch v1
Comment on attachment 526222 [details] [diff] [review] Patch v1 thank you for the catch!
Do we want this for Aurora?
What's the risk/reward for taking this? If this is super safe and the reward is high, please nominate the patch.
it's super safe. Not sure about reward though, I should know AT internals dor that. Under certain circumstances the deepest child is returned instead the direct child. Marco?
It would account for some inconsistencies I've seen with NVDA. But I've been running with the Nightly for a few days and didn#t notice any difference so far. But I'm nominating it anyway since this will improve things in some instances, and this is a regression introduced in FX4 IIRC.
Comment on attachment 526222 [details] [diff] [review] Patch v1 Nominating the patch, see previous comment.
Comment on attachment 526222 [details] [diff] [review] Patch v1 Heh. A=me
I took the liberty of landing this on Ms2ger's behalf on Aurora: http://hg.mozilla.org/mozilla-aurora/rev/389bc8414daf
The fix (comparison instead of assignment for the variables aWhichChild and eDeepestChild) is visible in the hg for latest beta build Setting this as Verified for Firefox 6 Beta: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:6.0) Gecko/20100101 Firefox/6.0