388744 - Off-by-one error in Windows drawing with inline-block and float

Status: RESOLVED FIXED

(Core :: Layout, defect)

Description

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: minimize testcase?

My z-index reftest fails on Windows but passes on Mac and Linux. The reference page is incorrect (see URL): some of the boxes are one pixel below than the top of their container, where they should be adjacent to the top of the container.

The reference page is quite simple. There's an absolutely positioned 'container' DIV with specified width and height. It has a blue 'negative' DIV child which is drawn in the wrong location; this is nothing but a regular DIV with specified width and height (and a negative bottom margin, but I don't see how that could affect its own rendering). The other parts of the test presumably can't affect the painting of 'negative'. I've attached the "minimized testcase" consisting of just 'container' and 'negative' --- but I haven't tested it on Windows so I don't know if it's a genuine testcase for this bug.

Comment 1

[Eli Friedman]

Hmm, the test certainly doesn't pass in my build, but the reference page looks fine.

Comment 2

[Eli Friedman]

Oops, the test doesn't pass in my build because I haven't updated to trunk recently enough.

I think the part that's causing the issue is the inline-block; I'll try to minimize it a bit more.

Comment 3

[Eli Friedman]

Comment on attachment 272939 [details]
minimize testcase?

This minimization doesn't reproduce the problem; the issue involves the interaction of the float and the inline-block.

Comment 4

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(away: 6/25-6/28)]

This is fixed by bug 421353. (now, the reftest is passed on Win32 too.)
If there are some remaining issues, please reopen this or file a new bug.

-> FIXED