NEW 139757
Delegates scrolling by wheel event should be work within contents area.
https://bugs.webkit.org/show_bug.cgi?id=139757
Summary Delegates scrolling by wheel event should be work within contents area.
Sanghyup Lee
Reported 2014-12-17 16:35:08 PST
Currently scroll position which changed by mouse wheel can be negative or exceed maximum scroll position when fixed layout is enabled.
Attachments
Patch (1.72 KB, patch)
2014-12-18 03:34 PST, Sanghyup Lee
no flags
Patch (5.08 KB, patch)
2015-02-11 22:48 PST, Sanghyup Lee
no flags
Patch (5.04 KB, patch)
2015-02-11 22:56 PST, Sanghyup Lee
no flags
Archive of layout-test-results from ews103 for mac-mavericks (523.39 KB, application/zip)
2015-02-11 23:43 PST, Build Bot
no flags
Archive of layout-test-results from ews106 for mac-mavericks-wk2 (667.48 KB, application/zip)
2015-02-11 23:49 PST, Build Bot
no flags
Patch (5.07 KB, patch)
2015-02-14 05:07 PST, Sanghyup Lee
no flags
Archive of layout-test-results from ews100 for mac-mavericks (551.56 KB, application/zip)
2015-02-14 05:52 PST, Build Bot
no flags
Patch (5.13 KB, patch)
2015-05-09 17:48 PDT, Sanghyup Lee
no flags
Archive of layout-test-results from ews100 for mac-mavericks (535.34 KB, application/zip)
2015-05-09 18:36 PDT, Build Bot
no flags
Patch (5.95 KB, patch)
2015-05-11 03:50 PDT, Sanghyup Lee
darin: review-
Sanghyup Lee
Comment 1 2014-12-18 03:34:42 PST
Darin Adler
Comment 2 2014-12-18 09:22:36 PST
Comment on attachment 243488 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=243488&action=review A bug fix like this needs a regression test. Could you please write a regression test that demonstrates the bug is fixed and add that to this patch? Or if that’s not possible, you could document why a regression test was impossible or impractical to make in the change log. > Source/WebCore/ChangeLog:3 > + [EFL] Delegates scrolling by wheel event should be work within contents area. This patch should affect any platform that uses delegatesScrolling, not just EFL, so it’s not appropriate for the bug title to say [EFL]. > Source/WebCore/page/FrameView.cpp:4351 > + ScrollView::setScrollOffset(IntPoint(newOffset)); Should call setScrollOffset, not ScrollView::setScrollOffset.
Sanghyup Lee
Comment 3 2014-12-18 16:46:58 PST
(In reply to comment #2) > Comment on attachment 243488 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=243488&action=review > > A bug fix like this needs a regression test. Could you please write a > regression test that demonstrates the bug is fixed and add that to this > patch? Or if that’s not possible, you could document why a regression test > was impossible or impractical to make in the change log. > > > Source/WebCore/ChangeLog:3 > > + [EFL] Delegates scrolling by wheel event should be work within contents area. > > This patch should affect any platform that uses delegatesScrolling, not just > EFL, so it’s not appropriate for the bug title to say [EFL]. > > > Source/WebCore/page/FrameView.cpp:4351 > > + ScrollView::setScrollOffset(IntPoint(newOffset)); > > Should call setScrollOffset, not ScrollView::setScrollOffset. Thank you for your review. I'll fix it.
Sanghyup Lee
Comment 4 2015-02-11 22:48:32 PST
Sanghyup Lee
Comment 5 2015-02-11 22:56:40 PST
Build Bot
Comment 6 2015-02-11 23:43:20 PST
Comment on attachment 246429 [details] Patch Attachment 246429 [details] did not pass mac-ews (mac): Output: http://webkit-queues.appspot.com/results/6325466193461248 New failing tests: fast/events/scroll-delegates-wheelevent.html
Build Bot
Comment 7 2015-02-11 23:43:23 PST
Created attachment 246434 [details] Archive of layout-test-results from ews103 for mac-mavericks The attached test failures were seen while running run-webkit-tests on the mac-ews. Bot: ews103 Port: mac-mavericks Platform: Mac OS X 10.9.5
Build Bot
Comment 8 2015-02-11 23:49:36 PST
Comment on attachment 246429 [details] Patch Attachment 246429 [details] did not pass mac-wk2-ews (mac-wk2): Output: http://webkit-queues.appspot.com/results/5180089683673088 New failing tests: fast/events/scroll-delegates-wheelevent.html
Build Bot
Comment 9 2015-02-11 23:49:39 PST
Created attachment 246435 [details] Archive of layout-test-results from ews106 for mac-mavericks-wk2 The attached test failures were seen while running run-webkit-tests on the mac-wk2-ews. Bot: ews106 Port: mac-mavericks-wk2 Platform: Mac OS X 10.9.5
Sanghyup Lee
Comment 10 2015-02-14 05:07:58 PST
Build Bot
Comment 11 2015-02-14 05:52:54 PST
Comment on attachment 246588 [details] Patch Attachment 246588 [details] did not pass mac-ews (mac): Output: http://webkit-queues.appspot.com/results/4826867815153664 New failing tests: fast/events/scroll-delegates-wheelevent.html
Build Bot
Comment 12 2015-02-14 05:52:58 PST
Created attachment 246589 [details] Archive of layout-test-results from ews100 for mac-mavericks The attached test failures were seen while running run-webkit-tests on the mac-ews. Bot: ews100 Port: mac-mavericks Platform: Mac OS X 10.9.5
Csaba Osztrogonác
Comment 13 2015-02-16 07:30:04 PST
Comment on attachment 246588 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=246588&action=review Is the Mac failure real or false positive? > Source/WebCore/ChangeLog:4 > + [WK2] Delegates scrolling by wheel event should be work within contents area. > + https://bugs.webkit.org/show_bug.cgi?id=139757 Why WK2, when you touch only one source which isn't in WebKit2 - Source/WebCore/page/FrameView.cpp ?
Sanghyup Lee
Comment 14 2015-02-16 17:10:22 PST
(In reply to comment #13) > Comment on attachment 246588 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=246588&action=review > > Is the Mac failure real or false positive? After check it, I'll leave comment. > > > Source/WebCore/ChangeLog:4 > > + [WK2] Delegates scrolling by wheel event should be work within contents area. > > + https://bugs.webkit.org/show_bug.cgi?id=139757 > > Why WK2, when you touch only one source which isn't in WebKit2 - > Source/WebCore/page/FrameView.cpp ? I Fixed it.
Sanghyup Lee
Comment 15 2015-05-09 17:48:00 PDT
Build Bot
Comment 16 2015-05-09 18:36:39 PDT
Comment on attachment 252790 [details] Patch Attachment 252790 [details] did not pass mac-ews (mac): Output: http://webkit-queues.appspot.com/results/5017664657293312 New failing tests: fast/events/scroll-delegates-wheelevent.html
Build Bot
Comment 17 2015-05-09 18:36:43 PDT
Created attachment 252793 [details] Archive of layout-test-results from ews100 for mac-mavericks The attached test failures were seen while running run-webkit-tests on the mac-ews. Bot: ews100 Port: mac-mavericks Platform: Mac OS X 10.9.5
Darin Adler
Comment 18 2015-05-10 13:53:08 PDT
Comment on attachment 252790 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=252790&action=review > Source/WebCore/page/FrameView.cpp:4460 > IntPoint oldPosition = scrollPosition(); > IntSize newOffset = IntSize(offset.width() - wheelEvent.deltaX(), offset.height() - wheelEvent.deltaY()); > if (offset != newOffset) { > - ScrollView::scrollTo(newOffset); > + setScrollOffset(IntPoint(newOffset)); > scrollPositionChanged(oldPosition, scrollPosition()); > didChangeScrollOffset(); > } With this patch as posted for review, scrollPositionChanged and didChangeScrollOffset will be called twice. Instead of these 8 lines of code, I suggest this one line of code: setScrollOffset(scrollPosition() + IntSize(-wheelEvent.deltaX(), -wheelEvent.deltaY())); I didn’t look at the failure on Mac; the one to that is running tests in this mode returned a failure.
Sanghyup Lee
Comment 19 2015-05-10 20:43:48 PDT
(In reply to comment #18) > Comment on attachment 252790 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=252790&action=review > > > Source/WebCore/page/FrameView.cpp:4460 > > IntPoint oldPosition = scrollPosition(); > > IntSize newOffset = IntSize(offset.width() - wheelEvent.deltaX(), offset.height() - wheelEvent.deltaY()); > > if (offset != newOffset) { > > - ScrollView::scrollTo(newOffset); > > + setScrollOffset(IntPoint(newOffset)); > > scrollPositionChanged(oldPosition, scrollPosition()); > > didChangeScrollOffset(); > > } > > With this patch as posted for review, scrollPositionChanged and > didChangeScrollOffset will be called twice. Instead of these 8 lines of > code, I suggest this one line of code: > > setScrollOffset(scrollPosition() + IntSize(-wheelEvent.deltaX(), > -wheelEvent.deltaY())); > > I didn’t look at the failure on Mac; the one to that is running tests in > this mode returned a failure. Thank you for your kind review and I'll fix it. As far as I know, delegate scrolling is for WK2. So Can I skip this test on Mac?
Sanghyup Lee
Comment 20 2015-05-11 03:50:17 PDT
Darin Adler
Comment 21 2015-05-11 09:54:01 PDT
Comment on attachment 252854 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=252854&action=review review- because skipping this test on Mac only isn’t right. But we can’t see that problem with EWS because currently only Mac EWS runs tests, so skipping on Mac is enough to make EWS happy. Someone needs to test this on iOS to see that it works as expected there. Could cause problems there because the scrolling UI involves rubber banding; this might interfere with that, maybe not. > LayoutTests/platform/mac/TestExpectations:287 > +# This port doesn't support delegate scrolling > +fast/events/scroll-delegates-wheelevent.html [ WontFix ] This doesn’t seem right. Mac is not the only port that has this issue. It’s just that the EWS bot only runs the test on Mac. So skipping this only for Mac is incorrect. I bet this also doesn’t work on Windows, and perhaps some other platforms as well. As far as I can tell, the “delegates scrolling” code path is used on iOS, and on platforms that use WebKit2 with COORDINATED_GRAPHICS. We shouldn’t be testing it on any other platforms unless we expect it to work. So just skipping this on Mac is not right.
Darin Adler
Comment 22 2015-05-11 09:54:19 PDT
Maybe there’s no wheel event on iOS?
Darin Adler
Comment 23 2015-05-11 09:54:50 PDT
“delegates scrolling” is used in both WebKit 1 and WebKit 2 on iOS.
Note You need to log in before you can comment on or make changes to this bug.