1918789 - Firefox hides a scrollbar (unlike other browsers) when body has `display:table` and either `overflow-x:hidden` or `overflow-y:hidden`.

Status: NEW

(Core :: Layout: Scrolling and Overflow, defect)

Description

[Daniel Holbert [:dholbert]]

Attachment: testcase 1 (with 'overflow-x: hidden')

[Spinning this off from bug 1895994 to serve as a platform bug, so bug 1895994 can track the site-report / site-patch that landed over here.]

STR:

1. Load attached testcase.

ACTUAL RESULTS:
No horizontal scrollbar on the window (you can't scroll to the right edge).

EXPECTED RESULTS:
Maybe there should be a horizontal scrollbar? All other browser engines (including pre-Chromium EdgeHTML) seem to show one, per bug 1895994 comment 15.

We've got the same bug (or difference-vs-other-browsers) with a missing vertical scrollbar if I use overflow-y:hidden in the testcase (rather than overflow-x).

However, if I specify hidden in both axes (overflow:hidden), then browsers agree that we should hide both scrollbars.

Comment 1

[Daniel Holbert [:dholbert]]

Attachment: testcase 2 (with 'overflow-y: hidden')

Comment 2

[Daniel Holbert [:dholbert]]

Attachment: testcase 3 (with overflow:hidden) - all browser agree on hiding both scrollbars here

Comment 3

[Daniel Holbert [:dholbert]]

emilio, do you happen to know where Chromium's body-to-root-scroller scrollbar-propagation code lives? I wonder if they've got some sort of weird special-case for this scenario that we can find and consider implementing? (Might conceivably be there for legacy/webcompat purposes, or due to something subtle related to the table-wrapper-box; not sure)

Comment 4

[Emilio Cobos Álvarez (:emilio)]

There are some suspicious LayoutBlock checks in here and here... Note that in Blink LayoutBlock is not only used for display: block but also for a bunch of other things... But notably not table.

I suspect it's just a bug in WebKit that Blink ended up inheriting. We should add this to Jeff's list of "bugs in other browsers that are worth fixing".

Comment 5

[Daniel Holbert [:dholbert]]

(In reply to Emilio Cobos Álvarez (:emilio) from comment #4)

I suspect it's just a bug in WebKit that Blink ended up inheriting.

That was my initial assumption, too -- though FWIW, IE and EdgeHTML match WebKit/Blink here, too. So maybe it was originally an IE quirk that WebKit matched, whether intentionally or accidentally? :)

(Opera 12 Presto matches Gecko on this, however, so we're not entirely the only engine that did what-seems-to-be-intuitive here.)

We should add this to Jeff's list of "bugs in other browsers that are worth fixing".

Sounds good! CC'ing Jeff and some folks involved with interop stuff.

Comment 6

[Emilio Cobos Álvarez (:emilio)]

Rune mentioned that this is because they end up computing the table to overflow: visible due to this code, so the propagation doesn't happen. A bit of a WTF tbh.