Bug 84178 – Relax ScrollView::adjustScrollPositionWithinRange in case constrainsScrollingToContentEdge is false

• Status: RESOLVED
• Resolution: FIXED
• Component: Platform

[Antonio Gomes]

Right now, when a port has delegatesScrolling set to FALSE, and constrainsScrollingToContentEdge set to FALSE (i.e. it accepts overscrolled position), calling ScrollView::setScrollPosition with an overscrolled position still gets the position clamped to the content size edges.

We could relax ::adjustScrollPositionWithinRange in that sense.

[Antonio Gomes]

*** Bug 84176 has been marked as a duplicate of this bug. ***

[Anders Carlsson]

I think doing so would break a lot of assumptions inside WebCore. I tried doing something like this in ScrollView::updateScrollbars and it did not work well.

[Antonio Gomes]

(In reply to comment #2)
> I think doing so would break a lot of assumptions inside WebCore. I tried doing something like this in ScrollView::updateScrollbars and it did not work well.

Would you remember what kind of assumptions? We run with this since a while without problems...

If we do not have such a "relaxation" if user over scrolls a webpage, say to (0, -20) webcore would be tracking a scroll position of (0, 0).

[Antonio Gomes]

Note that ScrollView::setScrollOffset does something similar already:

340 void ScrollView::setScrollOffset(const IntPoint& offset)
341 {                                                                          342     int horizontalOffset = offset.x();
343     int verticalOffset = offset.y();
344     if (constrainsScrollingToContentEdge()) {
345         horizontalOffset = max(min(horizontalOffset, contentsWidth() - visibleWidth()), 0);
346         verticalOffset = max(min(verticalOffset, contentsHeight() - visibleHeight()), 0); 
347     }        
....

[Antonio Gomes]

Created attachment 137582 [details]
(wrong patch, please ignore) patch

Patch for discussion.

Call  sites of ScrollView::adjustScrollPositionWithinRange from ::setScrollPosition and ::updateScrollbars.

[Antonio Gomes]

Created attachment 137708 [details]
(committed r114770, r=atreat) patch - the right patch this time

[Antonio Gomes]

<tonikitoo> andersca, sir, would you have some minutes to talk about? https://bugs.webkit.org/show_bug.cgi?id=84178
<tonikitoo> (shame we can not talk in person during the summit :( )
<andersca>  tonikitoo: yes, I think I was wrong about my assumption
<andersca>  tonikitoo: if you have a Mac you can test if it breaks things by going to a page with a vertical scrollbar, scroll to the bottom
<andersca>  tonikitoo: then hit command-f to show the find banner and escape to hide it
<tonikitoo> scrollview::setscrollposition calls it I think
<andersca>  tonikitoo: and make sure that the behavior is the same even with your patch
...
<tonikitoo> andersca, just to let you know: I got it tested on a safari build with and without the patch, and behaviour looks the same
<andersca>  tonikitoo: excellent

[Adam Treat]

Comment on attachment 137708 [details]
(committed r114770, r=atreat) patch - the right patch this time

lgtm and tested against mac port with no regression so r+

[Antonio Gomes]

Comment on attachment 137708 [details]
(committed r114770, r=atreat) patch - the right patch this time

Committed http://trac.webkit.org/changeset/114770