Remove unnecessary includes from KURL.h

Remove the dependencies from headers where possible. The biggest win is tightening up KURL.h

Remove duplicated headers from platform/ and web/ directories where the implementation file includes the same dependencies as the corresponding primary header.

BUG=317231
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=163438

[lgombos, 2013-12-08 19:30:06]

Cleanup..

[Nico, 2013-12-09 01:55:01]

The "Remove the dependencies from headers where possible. The biggest win is
tightening up KURL.h" part is cool, I'd put that first in the CL description.

Removing includes from the cpp that are already in the h seems less useful – in
most cases, you've only removed includes in the cpp that _have_ to be in the
cpp, so it's no worse I suppose, but there isn't a huge advantage to removing
these either (no build time improvement or anything). I'm not sure that part is
useful.

CL lgtm as is (since you already did the work for both parts; I'm saying this
more for future changes.)

https://codereview.chromium.org/94113005/diff/1/Source/core/html/LinkRelAttri...
File Source/core/html/LinkRelAttributeTest.cpp (right):

https://codereview.chromium.org/94113005/diff/1/Source/core/html/LinkRelAttri...
Source/core/html/LinkRelAttributeTest.cpp:32: #include
"core/html/LinkRelAttribute.h"
Source/core/html/LinkRelAttribute.h should have a `class String;`, right?

(but you still need the CString.h include in this file of course)

https://codereview.chromium.org/94113005/diff/1/Source/platform/mhtml/MHTMLPa...
File Source/platform/mhtml/MHTMLParser.cpp (left):

https://codereview.chromium.org/94113005/diff/1/Source/platform/mhtml/MHTMLPa...
Source/platform/mhtml/MHTMLParser.cpp:47: #include "wtf/text/WTFString.h"
Is the WTFString include in MHTMLParser.h necessary, or could String be
forward-declared there (if so, keep this include)?

[lgombos, 2013-12-09 03:56:04]

On 2013/12/09 01:55:01, Nico wrote:

> Removing includes from the cpp that are already in the h seems less useful
– in
> most cases, you've only removed includes in the cpp that _have_ to be in the
> cpp, so it's no worse I suppose, but there isn't a huge advantage to removing
> these either (no build time improvement or anything). I'm not sure that part
is
> useful.

Agreed. I personally think removing the duplicated includes from .cpp improves
readability as it makes it more obvious what are the _additional_ dependencies
required to implement the interface. If we keep the dups, this is not obvious as
the list is the interface+implementation list even in the implementation files.

Also technically these includes are proven to be dead code (as they are between
if #0 gards), that we tend to remove.

Lastly I have a script that does the work - I am mostly done with the tree by
now with the exception of the binding/ directory.

[lgombos, 2013-12-09 03:58:04]

@Eric - can you help reviewing web/. Thanks !

[eseidel, 2013-12-09 06:46:22]

web lgtm.

As Nico says, removing includes from headers is all that matters.  the .cpp
changes are mostly noise IMO.

[commit-bot: I haz the power, 2013-12-09 14:29:26]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/l.gombos@samsung.com/94113005/40001

[commit-bot: I haz the power, 2013-12-09 15:47:42]

Change committed as 163438