Bug 109867 – incorrect repainting when a table has a transform

• Status: RESOLVED
• Resolution: FIXED
• Component: Tables

[arno.]

Created attachment 188434 [details]
testcase

Hi,
I have a table-like element (display: table). This element has a transform.
Inside I have a block element. When I modify both the width and the height of the element, it is not repainted correctly: the rectangular part right of the current shape is not painted.
I attach a testcase which add one pixel in width and one pixel in width a few times. The result should be a rectangle, but actually, a triangle like shape is painted on the page.

[arno.]

Actually, this happens computeRectForRepaint for the inner div does not take into account the transformation of its parent. When the parent is not a table-like element, the rectangle takes into account the transformation. The difference between is because layout state is disabled in the generic case, but is enabled with a table.

[arno.]

Created attachment 188456 [details]
patch proposal: disable layout state during layout of table (and tableSection,tableRow) render object

[arno.]

For other objects, LayoutState is also disabled when there is a reflection. I wonder if we should also do the same thing.

[Eric Seidel (no email)]

Mitz is your man for LayoutState.

[Geoffrey Garen]

Comment on attachment 188456 [details]
patch proposal: disable layout state during layout of table (and tableSection,tableRow) render object

Looks like we should check for reflection as well.

[arno.]

Created attachment 200376 [details]
updated patch; check for reflections as well

new test (checking for reflection) fails without the patch and passes with the patch

[Build Bot]

Comment on attachment 200376 [details]
updated patch; check for reflections as well

Attachment 200376 [details] did not pass mac-ews (mac):
Output: http://webkit-queues.appspot.com/results/357286

New failing tests:
fast/frames/crash-remove-iframe-during-object-beforeload.html

[Build Bot]

Created attachment 200380 [details]
Archive of layout-test-results from webkit-ews-03 for mac-mountainlion

The attached test failures were seen while running run-webkit-tests on the mac-ews.
Bot: webkit-ews-03  Port: mac-mountainlion  Platform: Mac OS X 10.8.2

[Simon Fraser (smfr)]

Comment on attachment 200376 [details]
updated patch; check for reflections as well

View in context: https://bugs.webkit.org/attachment.cgi?id=200376&action=review

> LayoutTests/fast/repaint/reflection-table-layout-expected.html:6
> +    <meta charset="utf-8">
> +
> +    <title>Repaint test for bug 109867</title>

You can remove these.

> LayoutTests/fast/repaint/reflection-table-layout.html:6
> +    <meta charset="utf-8">
> +
> +    <title>Repaint test for bug 109867</title>

Ditto.

[Build Bot]

Comment on attachment 200376 [details]
updated patch; check for reflections as well

Attachment 200376 [details] did not pass mac-wk2-ews (mac-wk2):
Output: http://webkit-queues.appspot.com/results/384350

New failing tests:
fast/frames/crash-remove-iframe-during-object-beforeload.html

[Build Bot]

Created attachment 200388 [details]
Archive of layout-test-results from webkit-ews-15 for mac-mountainlion-wk2

The attached test failures were seen while running run-webkit-tests on the mac-wk2-ews.
Bot: webkit-ews-15  Port: mac-mountainlion-wk2  Platform: Mac OS X 10.8.2

[arno.]

Created attachment 200435 [details]
same patch; title and meta tags removed in LayoutTests

[Simon Fraser (smfr)]

Comment on attachment 200435 [details]
same patch; title and meta tags removed in LayoutTests

View in context: https://bugs.webkit.org/attachment.cgi?id=200435&action=review

> LayoutTests/fast/repaint/transform-table-layout.html:25
> +    window.setTimeout(function() {
> +        div.style.width = '2px';

It would be nice to have a test that doesn't require a timeout. It might be better to make a test that dumps repaint rectangles, rather than using a ref test.

[arno.]

Created attachment 200505 [details]
updated patch

Looks like there is an helper function to test repaint bugs. This patch use those function

[Simon Fraser (smfr)]

Comment on attachment 200505 [details]
updated patch

Excellent!

[WebKit Commit Bot]

Comment on attachment 200505 [details]
updated patch

Clearing flags on attachment: 200505

Committed r149550: <http://trac.webkit.org/changeset/149550>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.