REGRESSION: divs positioned absolutely to the right behave oddly in button tags

• Component: CSS
• Status: RESOLVED FIXED

[Dimitri Bouniol]

In the following code:

<button>Click Me<div style="position: absolute; right: 5px">Now</div></button>

the div will be to the right of the left side of the button, instead of the right side.
This problem does not exist in non-nightly versions of Safari/Webkit.
At the webpage above, you can see the problem at work in the navigation bar: the arrows are supposed to be to the right of the buttons, not the left.

[David Kilzer (:ddkilzer)]

Created attachment 14253 [details]
Test case from Comment #0

This looks the same to me in a local debug build of WebKit r21170 with Safari 2.0.4 (419.3) on Mac OS X 10.4.9 (8P135) as it does on shipping WebKit and Firefox 2.0.0.3.

Dimitri, could you post a screen shot of what you're seeing?

[Dimitri Bouniol]

Created attachment 14254 [details]
Difference between Safari and nightly WebKit

The window is Safari and the second one is the nightly Webkit.
The > image is positioned with "position: absolute; right: -12px; width: 12px;", yet it is anchored to the left side of the button.

[David Kilzer (:ddkilzer)]

Confirmed with a local debug build of WebKit r21170 with Safari 2.0.4 (419.3) on Mac OS X 10.4.9 (8P135).

This is also a regression as shipping Safari renders it differently, although Attachment #14253 [details] from Comment #0 is not a correct reduction of the issue.

[David Kilzer (:ddkilzer)]

Created attachment 14255 [details]
Partial reduction

Note that if you remove the 'body' style, you get a slightly different rendering on ToT WebKit (items are shifted down one or two pixels) versus shipping Safari.

[Dimitri Bouniol]

Created attachment 14256 [details]
Test case

Very basic test case that shows the problem.
The "new" should be displayed to the right of the button.

[Dave Hyatt]

Yikes, this is a bad regression.

[Dave Hyatt]

Created attachment 14264 [details]
Someone meant inline flows like spans and not inline blocks too

Fallout from Sam's positioning cleanup a while back.

[Maciej Stachowiak]

Comment on attachment 14264 [details]
Someone meant inline flows like spans and not inline blocks too

r=me

[Dave Hyatt]

Fixed in r21181