Implement more of Flush properly. This adds support for waiting on the messag...

Implement more of Flush properly. This adds support for waiting on the message
loop so we can now enforce that there is only one pending Flush at a time.
Also removes the ability to have multiple flushes pending at the same time.

TEST=covered by ppapi unit test
BUG=none

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=49899

[brettw, 2010-06-15 00:56:52]

I'll also pull ppapi when I submit this.

[darin (slow to review), 2010-06-15 15:45:50]

http://codereview.chromium.org/2862002/diff/7002/11001
File chrome/renderer/pepper_plugin_delegate_impl.cc (right):

http://codereview.chromium.org/2862002/diff/7002/11001#newcode72
chrome/renderer/pepper_plugin_delegate_impl.cc:72: if
(active_instances_.find(*i) != active_instances_.end())
couldn't the memory address of the PluginInstance be re-used?  this is really
begging for a "mutation safe set" like ObserverList.  it looks like
PluginInstance is reference counted, so perhaps you should use a set of
scoped_refptrs?

http://codereview.chromium.org/2862002/diff/7002/11003
File chrome/renderer/render_view.cc (right):

http://codereview.chromium.org/2862002/diff/7002/11003#newcode4314
chrome/renderer/render_view.cc:4314: if
(current_oldstyle_pepper_plugins_.find(*i) !=
same deal as before.

[brettw, 2010-06-15 16:38:29]

Uploaded a new patch with DEPS change and a comment clarifying why what we're
doing is OK. I didn't modify the comment for the "oldstyle" case since I want to
delete it soon.

[darin (slow to review), 2010-06-15 17:12:13]

OK, LGTM

On Tue, Jun 15, 2010 at 9:38 AM, <brettw@chromium.org> wrote:

> Uploaded a new patch with DEPS change and a comment clarifying why what
> we're
> doing is OK. I didn't modify the comment for the "oldstyle" case since I
> want to
> delete it soon.
>
>
> http://codereview.chromium.org/2862002/show
>