Use a shared thread for media operations.

Use a shared thread for media operations.

This reduces the number of threads created per media element by one.

BUG=61293
R=acolwell@chromium.org, jamesr@chromium.org, jochen@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=199531

[scherkus (not reviewing), 2013-05-09 21:59:32]

acolwell: take a look at media-ish stuff

jochen: take a look at DRT / content_shell changes. My general feeling hacking
on this code can be captured by the following image:
http://media.tumblr.com/tumblr_me0qrvGwDP1r3z80e.jpg

... so feel free to make recommendations, etc.

https://codereview.chromium.org/15005005/diff/11001/content/shell/renderer/we...
File content/shell/renderer/webkit_test_runner.cc (right):

https://codereview.chromium.org/15005005/diff/11001/content/shell/renderer/we...
content/shell/renderer/webkit_test_runner.cc:533: if
(test_media_stream_client_->IsMediaStream(url)) {
jochen: I didn't see a strong reason to stick webkit_glue::CreateMediaPlayer()
in test_media_stream_client.h as opposed to inlining the code here --
webkit/media/ is still glue code and it removes a layer of indirection

then again, I'd much rather have us not override cWMP() in the first place but I
feel that's something we'll have to tackle later :\

[scherkus (not reviewing), 2013-05-09 22:00:04]

[+jochen for realz]

jochen: take a look at DRT / content_shell changes. My general feeling hacking
on this code can be captured by the following image:
http://media.tumblr.com/tumblr_me0qrvGwDP1r3z80e.jpg

... so feel free to make recommendations, etc.

[acolwell GONE FROM CHROMIUM, 2013-05-09 23:25:45]

LGTM

[jochen (gone - plz use gerrit), 2013-05-10 07:37:34]

https://chromiumcodereview.appspot.com/15005005/diff/11001/content/shell/rend...
File content/shell/renderer/webkit_test_runner.cc (right):

https://chromiumcodereview.appspot.com/15005005/diff/11001/content/shell/rend...
content/shell/renderer/webkit_test_runner.cc:548: media_thread_.Start();
why don't you use the RenderThread's media thread?

You can CONTENT_EXPORT the getter, add a method to
content/test/layouttest_support.cc to access it

[scherkus (not reviewing), 2013-05-10 18:27:06]

PTAL

https://codereview.chromium.org/15005005/diff/11001/content/shell/renderer/we...
File content/shell/renderer/webkit_test_runner.cc (right):

https://codereview.chromium.org/15005005/diff/11001/content/shell/renderer/we...
content/shell/renderer/webkit_test_runner.cc:548: media_thread_.Start();
On 2013/05/10 07:37:34, jochen wrote:
> why don't you use the RenderThread's media thread?
> 
> You can CONTENT_EXPORT the getter, add a method to
> content/test/layouttest_support.cc to access it

This was exactly the sort of plumbing I was looking for! Thanks!

[jochen (gone - plz use gerrit), 2013-05-10 18:40:39]

lgtm

[scherkus (not reviewing), 2013-05-10 19:09:14]

+jamesr for OWNERS src/webkit/*

[scherkus (not reviewing), 2013-05-10 19:09:46]

On 2013/05/10 19:09:14, scherkus wrote:
> +jamesr for OWNERS src/webkit/*

... and content/*

[jamesr, 2013-05-10 19:13:17]

lgtm

I wish we had a better way to keep track of our threads in general.  Moving this
one to content::RenderThreadImpl is definitely a step in the right direction.

[scherkus (not reviewing), 2013-05-10 19:20:15]

On 2013/05/10 19:13:17, jamesr wrote:
> lgtm
> 
> I wish we had a better way to keep track of our threads in general.  Moving
this
> one to content::RenderThreadImpl is definitely a step in the right direction.

Yeah there's a lot of hidden one-off threads lurking around.

It'd be fun to make the base::Thread() constructor private + audit what breaks
:)

[scherkus (not reviewing), 2013-05-10 20:47:56]

Committed patchset #2 manually as r199531 (presubmit successful).