Create renderers for ExtensionHosts one a time to avoid blocking the UI.

Create renderers for ExtensionHosts one at a time to avoid blocking the UI.

BUG=14040
TEST=Install a bunch of extensions with toolstrips, then restart Chrome. The UI should be responsive while the toolstrips are loading.

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=27137

[Erik does not do reviews, 2009-09-24 20:47:57]

http://codereview.chromium.org/220028/diff/3001/2006
File chrome/browser/extensions/extension_host.cc (right):

http://codereview.chromium.org/220028/diff/3001/2006#newcode75
Line 75: MessageLoop::current()->PostTask(FROM_HERE,
I think we're guaranteed that all message loop tasks will be cleaned up before
singletons are destroyed, but wouldn't it be better to use a ScopedTaskFactory
here instead?  I think this would also allow you to avoid the retain/release
magic below.

http://codereview.chromium.org/220028/diff/3001/2006#newcode100
Line 100: struct RunnableMethodTraits<ExtensionHost::ProcessCreationQueue> {
a comment explaining this magic would help readability here

http://codereview.chromium.org/220028/diff/3001/2009
File chrome/browser/views/extensions/extension_view.cc (right):

http://codereview.chromium.org/220028/diff/3001/2009#newcode89
Line 89: host_->CreateRenderViewSoon(view);
I assume there's a gtk and mac change that need to be made for this as well.

http://codereview.chromium.org/220028/diff/3001/2010
File chrome/test/startup/startup_test.cc (right):

http://codereview.chromium.org/220028/diff/3001/2010#newcode185
Line 185: SetUpWithExtensionsProfile("e50");
is there a new profile that needs to be checked in?

[Matt Perry, 2009-09-24 21:11:36]

http://codereview.chromium.org/220028/diff/3001/2006
File chrome/browser/extensions/extension_host.cc (right):

http://codereview.chromium.org/220028/diff/3001/2006#newcode75
Line 75: MessageLoop::current()->PostTask(FROM_HERE,
On 2009/09/24 20:47:57, Erik Kay wrote:
> I think we're guaranteed that all message loop tasks will be cleaned up before
> singletons are destroyed, but wouldn't it be better to use a ScopedTaskFactory
> here instead?  I think this would also allow you to avoid the retain/release
> magic below.

Done.

http://codereview.chromium.org/220028/diff/3001/2009
File chrome/browser/views/extensions/extension_view.cc (right):

http://codereview.chromium.org/220028/diff/3001/2009#newcode89
Line 89: host_->CreateRenderViewSoon(view);
On 2009/09/24 20:47:57, Erik Kay wrote:
> I assume there's a gtk and mac change that need to be made for this as well.

Done.

http://codereview.chromium.org/220028/diff/3001/2010
File chrome/test/startup/startup_test.cc (right):

http://codereview.chromium.org/220028/diff/3001/2010#newcode185
Line 185: SetUpWithExtensionsProfile("e50");
On 2009/09/24 20:47:57, Erik Kay wrote:
> is there a new profile that needs to be checked in?

Oops, this was just for debugging. Reverted.

[Erik does not do reviews, 2009-09-24 21:41:10]

LGTM