Move UtilityProcessHost to content and move the message sending/dispatching to the clients. This...

Move UtilityProcessHost to content and move the message sending/dispatching to the clients.  This allows the content layer to use the class.

BUG=76697
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=88586

[Matt Perry, 2011-06-09 01:40:33]

LGTM modulo nits

http://codereview.chromium.org/6995095/diff/5003/chrome/utility/chrome_conten...
File chrome/utility/chrome_content_utility_client.cc (right):

http://codereview.chromium.org/6995095/diff/5003/chrome/utility/chrome_conten...
chrome/utility/chrome_content_utility_client.cc:1: // Copyright (c) 2011 The
Chromium Authors. All rights reserved.
I only skimmed this file, assuming most of it was cut+paste from UtilityThread.

http://codereview.chromium.org/6995095/diff/5003/chrome/utility/chrome_conten...
chrome/utility/chrome_content_utility_client.cc:90:
*unpacker.parsed_manifest()));
indent +1

http://codereview.chromium.org/6995095/diff/5003/chrome/utility/chrome_conten...
File chrome/utility/chrome_content_utility_client.h (right):

http://codereview.chromium.org/6995095/diff/5003/chrome/utility/chrome_conten...
chrome/utility/chrome_content_utility_client.h:54: base::PlatformFile pdf_file,
indent +2

http://codereview.chromium.org/6995095/diff/5003/content/browser/utility_proc...
File content/browser/utility_process_host.cc (right):

http://codereview.chromium.org/6995095/diff/5003/content/browser/utility_proc...
content/browser/utility_process_host.cc:42: if (!StartProcess())
It's not obvious that Send can only be called once per UPH instance. Could you
change it so that StartProcess does nothing if it's already been called?

[jam, 2011-06-09 02:45:57]

all done

http://codereview.chromium.org/6995095/diff/5003/chrome/utility/chrome_conten...
File chrome/utility/chrome_content_utility_client.cc (right):

http://codereview.chromium.org/6995095/diff/5003/chrome/utility/chrome_conten...
chrome/utility/chrome_content_utility_client.cc:1: // Copyright (c) 2011 The
Chromium Authors. All rights reserved.
On 2011/06/09 01:40:33, Matt Perry wrote:
> I only skimmed this file, assuming most of it was cut+paste from
UtilityThread.

yep all moved

[Matt Perry, 2011-06-09 20:54:48]

LGTM

http://codereview.chromium.org/6995095/diff/6010/content/browser/utility_proc...
File content/browser/utility_process_host.cc (right):

http://codereview.chromium.org/6995095/diff/6010/content/browser/utility_proc...
content/browser/utility_process_host.cc:69: started_ = true;
we should only set this at the end of the method, before returning true.

[jam, 2011-06-09 21:13:45]

http://codereview.chromium.org/6995095/diff/6010/content/browser/utility_proc...
File content/browser/utility_process_host.cc (right):

http://codereview.chromium.org/6995095/diff/6010/content/browser/utility_proc...
content/browser/utility_process_host.cc:69: started_ = true;
On 2011/06/09 20:54:49, Matt Perry wrote:
> we should only set this at the end of the method, before returning true.

I think it doesn't matter either way, because if it fails once, no point in keep
retrying (and this was similar to the old code)

but, more importantly, it turns out that the code below also caused messages to
be sent, which would call StartProcess in the overriden Send, so it has to be at
the top