159059 - [FIX]Border and hidden overflow handled incorrectly

Status: RESOLVED FIXED

(Core :: Layout, defect, P1)

Description

[David Hyatt]

If a block has a specified width of 0px, and if it has an overflowing block 
child or an overflowing line box, the block is incorrectly grown to enclose 
the overflowing block or line box.

If the block has a width > 0px, then the right thing happens.

Test case forthcoming.

Comment 1

[David Hyatt]

Attachment: A testcase for the bug.

Here's a testcase.

Comment 2

[David Hyatt]

Opera on Win32 does what I expect with the first testcase in the HTML file I 
attached, unlike what MOzilla seems to be doing.

Comment 3

[David Hyatt]

Relevant spec section:

http://www.w3.org/TR/REC-CSS2/visufx.html#overflow

Comment 4

[Boris Zbarsky [:bzbarsky]]

Attachment: Proposed patch

David, what's the deal with that CSS1 section that we're supposed to be
following here?

Comment 5

[David Hyatt]

Boy have I found a fun little monster.

Matrix of results on the third test case:

(1) WinIE - Expands red block to enclose blue, green and blue blocks correct
(2) Opera - Green block is not displayed! red and blue blocks correct
(3) Mozilla - Expands red block to enclose blue, green and blue blocks correct
(4) Mac IE - Expands red block to width of browser window(!!), 
    green block is also width of browser window, blue block correct
(5) Konqueror - Nothing shows up, width of 0 is special-cased to collapse down 
to nothing!

All browsers pass the width:1px test case with flying colors except for WinIE, 
which always expands the red block.

Comment 6

[Boris Zbarsky [:bzbarsky]]

The relevant CSS1 quote:

  The 'width' has a non-negative UA-defined minimum value (which may vary from
  element to element and even depend on other properties). If 'width' goes below
  this limit, either because it was set explicitly, or because it was 'auto' and
  the rules below would make it too small, the value will be replaced with the
  minimum value instead.

Note that we _only_ apply this when "width:0" is set, which means we aren't
really following CSS1 on this anyway....Also, as glazou points out CSS1 has no
concept of overflow, which is why this was probably necessary.

Comment 7

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

I think kipp was confused when he wrote that.  However, this bug is a duplicate
of bug 12750, which I filed almost three years ago.

Comment 8

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Comment on attachment 92546 [details] [diff] [review]
Proposed patch

This patch makes our behavior better than it is now.  r or sr = dbaron

Comment 9

[David Hyatt]

Comment on attachment 92546 [details] [diff] [review]
Proposed patch

r=hyatt

Comment 10

[Boris Zbarsky [:bzbarsky]]

Might as well take this...

Comment 11

[Boris Zbarsky [:bzbarsky]]

fix checked in.