Moved common parts of ChildProcessHost into chrome/common and created a Brows...

Moved common parts of ChildProcessHost into chrome/common and created a BrowserChildProcessHost with browser-specific implementation. This is in preparation of creating a ServiceChildProcessHost.
BUG=None
TEST=Test for regressions .


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=51593

[darin (slow to review), 2010-07-02 21:19:27]

LGTM

http://codereview.chromium.org/2885017/diff/14004/48006
File chrome/browser/browser_child_process_host.h (right):

http://codereview.chromium.org/2885017/diff/14004/48006#newcode37
chrome/browser/browser_child_process_host.h:37: virtual bool Send(IPC::Message*
msg);
OK, I see that it is best for ChildProcessHost to not implement
IPC::Message::Sender since the Send method would collide with this one.  Perhaps
ChildProcessHost::SendOnChannel should be protected though?

http://codereview.chromium.org/2885017/diff/14004/48039
File chrome/common/child_process_host.h (right):

http://codereview.chromium.org/2885017/diff/14004/48039#newcode24
chrome/common/child_process_host.h:24: // Subclasses are responsible for the
actual launching and terminating of the
nit: start this second sentence on the same line as the previous.

http://codereview.chromium.org/2885017/diff/14004/48039#newcode47
chrome/common/child_process_host.h:47: bool SendOnChannel(IPC::Message* msg);
Please document the behavior of this method.  I'm guessing it behaves just like
IPC::Message::Sender::Send.  Should ChildProcessHost just implement the Sender
interface?

[sanjeevr, 2010-07-02 21:41:56]

Made all the changes.

http://codereview.chromium.org/2885017/diff/14004/48006
File chrome/browser/browser_child_process_host.h (right):

http://codereview.chromium.org/2885017/diff/14004/48006#newcode37
chrome/browser/browser_child_process_host.h:37: virtual bool Send(IPC::Message*
msg);
On 2010/07/02 21:19:27, darin wrote:
> OK, I see that it is best for ChildProcessHost to not implement
> IPC::Message::Sender since the Send method would collide with this one. 
Perhaps
> ChildProcessHost::SendOnChannel should be protected though?

Done.

http://codereview.chromium.org/2885017/diff/14004/48039
File chrome/common/child_process_host.h (right):

http://codereview.chromium.org/2885017/diff/14004/48039#newcode24
chrome/common/child_process_host.h:24: // Subclasses are responsible for the
actual launching and terminating of the
On 2010/07/02 21:19:27, darin wrote:
> nit: start this second sentence on the same line as the previous.

Done.

http://codereview.chromium.org/2885017/diff/14004/48039#newcode47
chrome/common/child_process_host.h:47: bool SendOnChannel(IPC::Message* msg);
On 2010/07/02 21:19:27, darin wrote:
> Please document the behavior of this method.  I'm guessing it behaves just
like
> IPC::Message::Sender::Send.  Should ChildProcessHost just implement the Sender
> interface?

Done.

[darin (slow to review), 2010-07-02 23:50:53]

LGTM!