943459 - Add flexbox reftests with negative margins on flex items, and with auto margins in an overconstrained flexbox

Status: NEW

(Core :: Layout, defect)

Description

[Daniel Holbert [:dholbert]]

mats points out in bug 939896 comment 26 that we don't seem to have reftests for flexbox with negative margins on flex items.

Filing this bug on adding some.

Comment 1

[Daniel Holbert [:dholbert]]

(Broadening this to also cover testing auto margins in flex containers with not enough space to assign any to the auto margin, too - the other thing that mats suggested testing in that bug comment.)

Comment 2

[Daniel Holbert [:dholbert]]

FWIW, existing tests for margins on flex items are listed here:
http://mxr.mozilla.org/mozilla-central/source/layout/reftests/w3c-css/submitted/flexbox/
They're named flexbox-margin-* and flexbox-mbp-*.

Comment 3

[Guido Bouman]

I would like to point out a bug in the flexbox implementation that has to do with exactly these missing tests.

See http://codepen.io/anon/pen/WbjZdp?editors=110

Setup:
Container div with padding, overflow-y: auto & display: flex;
Content div with margin: auto;

Problem:
When the content is larger than the container, you're able to scroll, but the bottom padding on the container disappears.

Works well on Chrome, but the bug also appears on Safari.

Should I submit a new bug for this one, or will this be picked up once these tests are implemented?

Comment 4

[Daniel Holbert [:dholbert]]

I don't see that as being related to this bug -- it doesn't involve any negative margins.  So -- yes, please do file a new bug, and CC me. (I'm working to reduce your testcase right now, to figure out what's making us disagree with Chrome on the rendering there.)

Comment 5

[Daniel Holbert [:dholbert]]

Ah, it looks like your issue is just that the "padding" on an 'overflow:scroll' element is not reachable via scrolling. That doesn't have anything to do with flexbox, and we're actually following the spec -- see bug 748518.  Blink/WebKit are wrong (according to the spec) on this.

Comment 6

[Guido Bouman]

My submission was mainly in relation to the part: "and with auto margins in an overconstrained flexbox".

But apparently this is according to the spec. Why the spec would state that is a freaking mystery to me, but that's something I'd have to bring up with the working group...