Adding tracking of HWND creation/destruction

This CL adds macro used to track the creation and destruction
of HWNDs, in an attempt to detect potential double-delete.
A double-delete of a HWND might be responsible for the crasher
http://crbug.com/4714




Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=9109

[sky, 2009-02-03 18:02:04]

Is it worth tracking the other HWNDs we create? Tables, trees, buttons ...

If you want to run this on the buildbot, I think you'll need to hide behind a
switch.

http://codereview.chromium.org/21018/diff/1007/1010
File base/win_util.cc (right):

http://codereview.chromium.org/21018/diff/1007/1010#newcode402
Line 402: "how you got that assertion.";
Does the Location that was used in the creation have anything useful that should
be logged here?

[jcampan, 2009-02-03 19:32:42]

Modified NativeControl to track its HWND, which covers buttons, tab panned and
the rest.
As a result, NativeControl now always has to subclass its HWND (it only used to
do it for on key down notifications).

I am keeping it as a DCHECK. Later tonight I'll make it a check (tonight so
there is less a chance that I annoy anyone with possible false positives) and
watch the build bot cycle to see if it catches something.

http://codereview.chromium.org/21018/diff/1007/1010
File base/win_util.cc (right):

http://codereview.chromium.org/21018/diff/1007/1010#newcode402
Line 402: "how you got that assertion.";
On 2009/02/03 18:02:04, sky wrote:
> Does the Location that was used in the creation have anything useful that
should
> be logged here?
Good suggestion. Done.

[sky, 2009-02-03 20:37:24]

LGTM

http://codereview.chromium.org/21018/diff/42/45
File base/win_util.cc (right):

http://codereview.chromium.org/21018/diff/42/45#newcode412
Line 412: "Double delete of HWND 0x" << hwnd << "\n" <<
will "<< hwnd" give you hex, or decimal?