Add template parameter to ElementTraversal to iterate over Elements of a specific type

Add template parameter to ElementTraversal to iterate over Elements of a specific type

Add template parameter to ElementTraversal to iterate over Elements of a
specific type. The resulting code looks like:
HTMLTableRowElement* row = Traversal<HTMLTableRowElement>::firstWithin(table);

This avoids having to traverse generic Elements and then explicitly check their
tag name to make sure it is a specific Element subclass.

This CL only updates one file to use this functionality. The rest of the code
will be ported in a follow-up patch to reduce patch size.

This patch is based on the following WebKit revision by <antti@apple.com>;:
http://trac.webkit.org/changeset/154481

R=esprehn@chromium.org, eseidel
BUG=346733
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=168689

[esprehn, 2014-02-25 19:08:30]

How does the general form of Traversal<Element>::firstWithin() work? Is there an
isElement() method?

[Inactive, 2014-02-25 19:20:39]

On 2014/02/25 19:08:30, esprehn wrote:
> How does the general form of Traversal<Element>::firstWithin() work? Is there
an
> isElement() method?

First of all, you can use ElementTraversal::firstWithin() instead of
Traversal<Element>::firstWithin(). This is equivalent and still works.
The reason Traversal<Element> works is because of the following functions in
Element.h:
template <typename T> inline bool isElementOfType(const Node& node) { return
node.isElementNode() && isElementOfType<const T>(toElement(node)); }
template <> inline bool isElementOfType<const Element>(const Element&) { return
true; }

Those were added in my previous CL.

[esprehn, 2014-02-26 20:17:48]

https://codereview.chromium.org/179333004/diff/20001/Source/core/dom/ElementT...
File Source/core/dom/ElementTraversal.h (right):

https://codereview.chromium.org/179333004/diff/20001/Source/core/dom/ElementT...
Source/core/dom/ElementTraversal.h:105: template <class ElementType>
Why stack two template<> decls instead of doing template<class ElementType,
class NodeType> ?

[esprehn, 2014-02-26 20:17:53]

lgtm

[Inactive, 2014-02-26 20:24:20]

https://codereview.chromium.org/179333004/diff/20001/Source/core/dom/ElementT...
File Source/core/dom/ElementTraversal.h (right):

https://codereview.chromium.org/179333004/diff/20001/Source/core/dom/ElementT...
Source/core/dom/ElementTraversal.h:105: template <class ElementType>
On 2014/02/26 20:17:49, esprehn wrote:
> Why stack two template<> decls instead of doing template<class ElementType,
> class NodeType> ?

This does not build (just double-checked). one template is for the class, and
the other is for the method. It is a templated method/function inside a
templated class.

[Inactive, 2014-02-26 20:29:13]

https://codereview.chromium.org/179333004/diff/20001/Source/core/dom/ElementT...
File Source/core/dom/ElementTraversal.h (right):

https://codereview.chromium.org/179333004/diff/20001/Source/core/dom/ElementT...
Source/core/dom/ElementTraversal.h:105: template <class ElementType>
On 2014/02/26 20:17:49, esprehn wrote:
> Why stack two template<> decls instead of doing template<class ElementType,
> class NodeType> ?

FYI, since I googled it as well:
http://stackoverflow.com/questions/8640390/c-template-function-inside-templat...

[Inactive, 2014-02-26 20:29:22]

The CQ bit was checked by ch.dumez@samsung.com

[commit-bot: I haz the power, 2014-02-26 20:29:37]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/ch.dumez@samsung.com/179333004/40001

[Inactive, 2014-02-27 00:27:19]

Committed patchset #3 manually as r167945 (presubmit successful).

[Inactive, 2014-02-27 02:04:13]

A revert of this CL has been created in
https://codereview.chromium.org/182583002/ by ch.dumez@samsung.com.

The reason for reverting is: Does not work as intended, I noticed breakages when
porting more code to the new API. Reverting until I figure out the problem..

[Inactive, 2014-02-27 20:28:54]

I have fixed the patch. The issue was that the implementation of firstWithin()
was not doing the right thing for Element subclasses. firstWithin() for pure
Elements is basically a firstChild() as we know that the parent of an Element is
always an Element (or the root).

However, for Element subclasses, firstWithin() possibly traverses down to a
descendant and not simply a child.

I introduced firstChild() / lastChild() traversal methods. Those are equivalent
to firstWithin() / lastWithin() for pure Elements, but not for other types of
Elements (where we need the first *descendant*).

The new patch is very close to the WebKit implementation. I ported a bit more
code to the new traversal API to get a better test coverage this time. My
earlier patch did not cause any failures because the code I ported did not
require firstWithin() to go down to a descendant (the table row was a direct
child).

Please take another look.

[Inactive, 2014-03-04 14:57:03]

Esprehn, could you please take another look when you get a chance? Thanks.

[esprehn, 2014-03-06 19:31:53]

lgtm, thanks!

[Inactive, 2014-03-06 21:27:08]

The CQ bit was checked by ch.dumez@samsung.com

[commit-bot: I haz the power, 2014-03-06 21:27:29]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/ch.dumez@samsung.com/179333004/60001

[commit-bot: I haz the power, 2014-03-06 21:27:55]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-03-06 21:27:55]

Try jobs failed on following builders: linux_blink, win_blink_rel

[Inactive, 2014-03-06 22:05:40]

The CQ bit was checked by ch.dumez@samsung.com

[commit-bot: I haz the power, 2014-03-06 22:06:32]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/ch.dumez@samsung.com/179333004/60001

[commit-bot: I haz the power, 2014-03-07 00:29:29]

Change committed as 168689