[webkit-reviews] review canceled: [Bug 80808] absolutely-positioned element does not update width when container width changes : [Attachment 143018] Patch

[bugzilla-daemon at webkit.org]

Julien Chaffraix <jchaffraix at webkit.org> has canceled SravanKumar S(:sravan)
<ssandela at innominds.com>'s request for review:
Bug 80808: absolutely-positioned element does not update width when container
width changes
https://bugs.webkit.org/show_bug.cgi?id=80808

Attachment 143018: Patch
https://bugs.webkit.org/attachment.cgi?id=143018&action=review

------- Additional Comments from Julien Chaffraix <jchaffraix at webkit.org>
View in context: https://bugs.webkit.org/attachment.cgi?id=143018&action=review


> Source/WebCore/rendering/RenderBlock.cpp:2476
> +	       r->setChildNeedsLayout(true, MarkOnlyThis);

I have strong doubts that this is the right fix. Could you explain why you
believe we need this?

Looking at the issue, it's definitely not a repainting bug as selecting doesn't
repaint the outline as it should be. Also the text is properly laid out, only
the outline that is wrong. I would look into the bounding box / visual overflow
from ".abs" and try to understand why it is wrongly computed in this case (we
are likely not including the overflow from the new child but I don't know why).
You can also track the paintOutline function and backtrack the values from
there.

> LayoutTests/ChangeLog:12
> +	   *
compositing/absolute-position-change-width-with-inline-container-expected.html:
Added.
> +	   *
compositing/absolute-position-change-width-with-inline-container.html: Added.

Why is the test in compositing/ when it involves 0 composition?

>
LayoutTests/compositing/absolute-position-change-width-with-inline-container.ht
ml:26
> +	   <p>When you click <button id="more">moooore</button>, the span.abs
should expand with the container</p>
> +	   <p><span id="container"><span class="abs"> </span>Words, words,
words</span>

This test passes for me locally on Chromium (haven't tested under DRT). Also
can we have a test that doesn't involve a button that doesn't work? We also
need some mention of the bug id, title and condition for passing.