244232 2022-08-22 19:13:48 -0700 overscroll-behavior: none doesn't prevent overscroll when page is too small to scroll 2023-01-23 11:18:32 -0800 1 1 1 Unclassified WebKit Scrolling Other All All RESOLVED FIXED BrowserCompat, InRadar P2 Normal --- 1 krevis nmouchtaris karlcow nmouchtaris simon.fraser webkit-bug-importer oldest_to_newest 1893136 0 krevis 2022-08-22 19:13:48 -0700 When a document has overscroll-behavior: none, but the page's height is too small to be scrollable, the page can still be overscrolled using a touch (iOS) or trackpad scroll (iOS, Mac). Steps: 1. Load attached html file in Safari Technical Preview 151 on macOS 12, or Safari on iOS/iPadOS 16 beta, or Safari in macOS 13 beta 2. On Mac, scroll up and down using two-finger scroll gesture on trackpad. 3. On iPhone, scroll up and down with touch. 4. On iPad, scroll up and down with touch, or with two-finger scroll gesture on trackpad, or with mouse wheel. Expected: Since the html and body elements have `overscroll-behavior: none`, overscroll aka "bouncing" should not happen. Actual: Overscroll aka "bouncing" does happen. If you click on the element that increases the body height, so scrolling is possible, then overscroll stops working. Notes: Both Chrome and Firefox behave as expected. (Chrome 104.0.5112.101 and Firefox 103.0.2 on macOS 12.5.1.) The Overscroll Behavior spec says, about overscroll-behavior: none, that "this element must also not perform local boundary default actions such as showing any overscroll affordances". https://drafts.csswg.org/css-overscroll/#valdef-overscroll-behavior-none Also, "If a scroll container has no potential to scroll, because it does not overflow in the direction of the scroll, the element is always considered to be at the scroll boundary." https://drafts.csswg.org/css-overscroll/#scroll-boundary Nothing in the spec appears to say that the overscroll-behavior should be ignored in this case. There's an existing bug with a similar complaint: https://bugs.webkit.org/show_bug.cgi?id=243452 but this case is simpler, with no nested scroll views, just a simple document. 1893137 1 461811 krevis 2022-08-22 19:15:13 -0700 Created attachment 461811 Test case 1893138 2 krevis 2022-08-22 19:19:52 -0700 I'm suspicious of this code in WebCore/page/FrameView.cpp: OverscrollBehavior FrameView::verticalOverscrollBehavior() const { auto* document = frame().document(); auto scrollingObject = document && document->documentElement() ? document->documentElement()->renderer() : nullptr; if (scrollingObject && renderView() && renderView()->canBeScrolledAndHasScrollableArea()) return scrollingObject->style().overscrollBehaviorY(); return OverscrollBehavior::Auto; } When renderView()->canBeScrolledAndHasScrollableArea() is false, we ignore the style and return Auto instead. What is that check trying to accomplish? 1893169 3 webkit-bug-importer 2022-08-23 00:10:32 -0700 <rdar://problem/99018981> 1926756 4 nmouchtaris 2023-01-19 16:40:42 -0800 Pull request: https://github.com/WebKit/WebKit/pull/8859 1927523 5 ews-feeder 2023-01-23 11:18:29 -0800 Committed 259227@main (487c9dfe271c): <https://commits.webkit.org/259227@main> Reviewed commits have been landed. Closing PR #8859 and removing active labels. 461811 2022-08-22 19:15:13 -0700 2022-08-22 19:15:13 -0700 Test case 244232.html text/html 1185 krevis PCFET0NUWVBFIGh0bWw+CjxodG1sIGxhbmc9ImVuIiA+CjxoZWFkPgogIDxzdHlsZT4KICAgIGh0 bWwgeyAvKiBGb3IgU2FmYXJpIGFuZCBGaXJlZm94ICovCiAgICAgIG92ZXJzY3JvbGwtYmVoYXZp b3I6IG5vbmU7CiAgICB9CiAgICBib2R5IHsgLyogRm9yIENocm9tZSAqLwogICAgICBvdmVyc2Ny b2xsLWJlaGF2aW9yOiBub25lOwogICAgfQogIDwvc3R5bGU+CiAgPG1ldGEgY2hhcnNldD0iVVRG LTgiPgogIDx0aXRsZT5vdmVyc2Nyb2xsLWJlaGF2aW9yOiBub25lIGRvZXNuJ3QgcHJldmVudCBv dmVyc2Nyb2xsIHdoZW4gcGFnZSBpcyB0b28gc21hbGwgdG8gc2Nyb2xsPC90aXRsZT4KPC9oZWFk Pgo8Ym9keT4KICA8aDI+PGEgaHJlZj0iaHR0cHM6Ly9idWdzLndlYmtpdC5vcmcvc2hvd19idWcu Y2dpP2lkPTI0NDIzMiI+b3ZlcnNjcm9sbC1iZWhhdmlvcjogbm9uZSBkb2Vzbid0IHByZXZlbnQg b3ZlcnNjcm9sbCB3aGVuIHBhZ2UgaXMgdG9vIHNtYWxsIHRvIHNjcm9sbDwvYT48L2gyPgogIDxw PlRyeSBzY3JvbGxpbmcgdXAgYW5kIGRvd24gd2l0aCB0b3VjaCBvciB0cmFja3BhZC48L3A+CiAg PHAgaWQ9InRvZ2dsZSI+PGI+VG91Y2ggb3IgY2xpY2sgaGVyZSB0byB0b2dnbGUgd2hldGhlciBw YWdlIGhlaWdodCBpcyBzY3JvbGxhYmxlIG9yIG5vdC48L2I+PC9wPgogIDxwPjxiPkV4cGVjdGVk OjwvYj4gU2luY2Ugb3ZlcnNjcm9sbC1iZWhhdmlvciBpcyBub25lLCB0aGUgcGFnZSBzaG91bGQg bmV2ZXIgb3ZlcnNjcm9sbC4KICA8YnI+PGI+QWN0dWFsOjwvYj4gSW4gU2FmYXJpLCB3aGVuIHRo ZSBwYWdlIGlzIHNob3J0IGFuZCBpcyBub3Qgc2Nyb2xsYWJsZSwgdGhlIHBhZ2Ugb3ZlcnNjcm9s bHMuPC9wPgo8L2JvZHk+CjxzY3JpcHQ+CiAgZnVuY3Rpb24gdG9nZ2xlSGVpZ2h0KCkgewogICAg aWYgKGRvY3VtZW50LmJvZHkuc3R5bGUuaGVpZ2h0ID09PSAiIikgewogICAgICBkb2N1bWVudC5i b2R5LnN0eWxlLmhlaWdodCA9ICI1MDAwcHgiOwogICAgfQogICAgZWxzZSB7CiAgICAgIGRvY3Vt ZW50LmJvZHkuc3R5bGUuaGVpZ2h0ID0gIiI7CiAgICB9CiAgfQoKICBkb2N1bWVudC5nZXRFbGVt ZW50QnlJZCgndG9nZ2xlJykuYWRkRXZlbnRMaXN0ZW5lcigncG9pbnRlcmRvd24nLCAoKSA9PiB7 CiAgICB0b2dnbGVIZWlnaHQoKTsKICB9KTsKPC9zY3JpcHQ+CjwvaHRtbD4K