If your mouse is in a frame that can scroll, and you mousewheel to the bottom of the frame, the scrolling action will continue to its parent (usually the document). However, if you are panscrolling, when you scroll to the bottom of the frame, the scrolling does not continue to scroll the parent, and just stops the scrolling when it is done scrolling the frame. I am not sure if this is by design or a bug, but I am filing it here to open up discussion on the proper behavior. Fixing this to continue panscrolling/panning to the parent would have 2 benefits: 1) It would make these ways of scrolling more like mousewheel scrolling. 2) It would make scrolling frames and iFrames functionally equivalent to scrolling divs with overflow. On the attached URL, go into one of the iFrames with scrollbars, and try mouse-wheeling and pan-scrolling to reproduce the behavior that is discussed here.
Created attachment 38999 [details] Fix I'm not sure if this is a behavior we want, the upsides are listed in the initial bug report.
What does IE do?
Hyatt should probably have a look at this, too.
IE doesn't jump out of frames or divs with overflow. Firefox doesn't even pan scroll divs with overflow and doesn't jump out of iFrames. Opera doesn't pan scroll divs with overflow either, and doesn't jump out of iFrames.
Comment on attachment 38999 [details] Fix Can't we write a LayoutTest for this? Seems we either need a test, or an explanation in the ChangeLog as to why testing is impossible.
Brian, planning to working on the layout test? If not, I can try to finish your work.
(In reply to comment #6) > Brian, planning to working on the layout test? If not, I can try to finish your work. There was some debate on whether or not this was the desired behavior, and if this bug should just be closed as RESOLVED/WONTFIX or something along those lines. This was a while ago, so I don't completely remember all that was discussed, but I can ask around and see if anyone else remembers anything. I wasn't planning to come back to this any time soon, so if you want to put together a layout test for it, that would be great. It should be doable.
> There was some debate on whether or not this was the desired behavior, and if this bug should just be closed as RESOLVED/WONTFIX or something along those lines. This was a while ago, so I don't completely remember all that was discussed, but I can ask around and see if anyone else remembers anything. Brian, please ask. I would gladly help with that if it is the way to go OR remove this fixed from the code and WONTFIX this bug. Thanks
I have been doing work in this area and wanted to know if this issue is still relevant.