CSS 2.1 failure: border-spacing-applies-to-015.htm

[Robert Hogan]

WebKit only allows one caption per table.

[Robert Hogan]

Created attachment 110553 [details]
Patch

[Dave Hyatt]

Comment on attachment 110553 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=110553&action=review

> Source/WebCore/rendering/RenderTable.cpp:203
> +        if (node())
> +          node()->setNeedsStyleRecalc();

Funny looking indentation here.

> Source/WebCore/rendering/RenderTable.cpp:738
> -                    recalcCaption(toRenderBlock(child));
> +                    toRenderBlock(child)->setNeedsLayout(true);

Why do you even need to mark it as needing layout? Can't you just yank this code completely?

[Robert Hogan]

Created attachment 113214 [details]
Patch

[Dave Hyatt]

Comment on attachment 113214 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=113214&action=review

> Source/WebCore/rendering/RenderTable.cpp:1196
> -            if (child->isBox() && !toRenderBox(child)->hasSelfPaintingLayer() && (child->isTableSection() || child == m_caption)) {
> +            if (child->isBox() && !toRenderBox(child)->hasSelfPaintingLayer() && (child->isTableSection() || m_captions.contains(child))) {

Seems like it would be faster instead of asking if the child is contained in a Vector (which has to crawl the whole vector) to just say:

(child->isTableSection() || (child->isRenderBlock() && child->style()->display() == TABLE_CAPTION))

[Dave Hyatt]

Comment on attachment 113214 [details]
Patch

r=me

[Robert Hogan]

Committed r100177: <http://trac.webkit.org/changeset/100177>