Remove ExtensionMessageService's legacy support for devtools.

Remove ExtensionMessageService's legacy support for devtools.
- Remove OpenSpecialChannelTo*.
- Make EMS non-refcounted now that Profile is the sole owner.
- Make MessagePorts hold a RenderProcessHost now that processes are the only
types of port. I also changed it to hold a process instead of a RenderViewHost,
which doesn't make a difference from an IPC messaging standpoint.

BUG=104625
TEST=no

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=126225

[Matt Perry, 2012-03-09 23:01:41]

Yoyo: please review
yurys, sdoyon, pfeldman: FYI. I figured thanks to
http://codereview.chromium.org/8635005/ that OpenSpecialChannel was no longer
needed. Please correct me if I'm mistaken.

[Yoyo Zhou, 2012-03-10 01:05:46]

On 2012/03/09 23:01:41, Matt Perry wrote:
> Yoyo: please review
> yurys, sdoyon, pfeldman: FYI. I figured thanks to
> http://codereview.chromium.org/8635005/ that OpenSpecialChannel was no longer
> needed. Please correct me if I'm mistaken.

Cool, but this will run into some merge conflicts with my work on
crbug.com/104095, which I'll hopefully finally get to stick this weekend.
(Notably, ExtensionSystem will own it rather than Profile.)

[Yoyo Zhou, 2012-03-10 01:36:54]

http://codereview.chromium.org/9667015/diff/1/chrome/browser/extensions/exten...
File chrome/browser/extensions/extension_message_service.cc (right):

http://codereview.chromium.org/9667015/diff/1/chrome/browser/extensions/exten...
chrome/browser/extensions/extension_message_service.cc:135:
ExtensionMessageService::ExtensionMessageService(Profile* profile) {
Apparently you no longer need this argument.

http://codereview.chromium.org/9667015/diff/1/chrome/browser/profiles/profile...
File chrome/browser/profiles/profile_impl.cc (right):

http://codereview.chromium.org/9667015/diff/1/chrome/browser/profiles/profile...
chrome/browser/profiles/profile_impl.cc:447:
extension_message_service_.reset(new ExtensionMessageService(this));
Yeah, these changes can basically be applied verbatim to extension_system.cc
once that gets in.

[Yoyo Zhou, 2012-03-12 18:33:51]

LGTM

https://chromiumcodereview.appspot.com/9667015/diff/1/chrome/browser/profiles...
File chrome/browser/profiles/profile_impl.cc (right):

https://chromiumcodereview.appspot.com/9667015/diff/1/chrome/browser/profiles...
chrome/browser/profiles/profile_impl.cc:447:
extension_message_service_.reset(new ExtensionMessageService(this));
On 2012/03/10 01:36:54, Yoyo Zhou wrote:
> Yeah, these changes can basically be applied verbatim to extension_system.cc
> once that gets in.

It looks like my change is actually going to be blocked for a while, so you can
go ahead with this.

[Matt Perry, 2012-03-12 21:06:33]

miranda: can I get an OWNERS lgtm for chrome/browser/profiles ?

[Miranda Callahan, 2012-03-12 21:07:49]

On 2012/03/12 21:06:33, Matt Perry wrote:
> miranda: can I get an OWNERS lgtm for chrome/browser/profiles ?

Yes, you can. OWNERS lgtm for profiles/*.

[Matt Perry, 2012-03-12 21:15:13]

Thanks!

[pfeldman, 2012-03-13 17:56:37]

Did you use the right bug for this? Note, that all this
extension_devtools_bridge business was created for the timeline / speedtracer
API. We are deprecating it and it'll be removed in M20.

[Matt Perry, 2012-03-13 19:17:39]

On 2012/03/13 17:56:37, pfeldman wrote:
> Did you use the right bug for this?

Do you know of a better bug to link here? I used the same one as what was used
in the removal of the extension_ports_service.

> Note, that all this
> extension_devtools_bridge business was created for the timeline / speedtracer
> API. We are deprecating it and it'll be removed in M20.

I'm not familiar enough with the devtools stuff to know what this means. Will it
have an impact on ExtensionMessageService?