Preload PNaCl plugin when the webapp is started.

Preload PNaCl plugin when the webapp is started.

Preloading the plugin starts NaCl translator, which saves time when
the user tries to connect.

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=290040

[Wez, 2014-08-12 22:03:31]

https://codereview.chromium.org/467903002/diff/1/remoting/webapp/client_plugi...
File remoting/webapp/client_plugin.js (right):

https://codereview.chromium.org/467903002/diff/1/remoting/webapp/client_plugi...
remoting/webapp/client_plugin.js:134:
document.body.appendChild(remoting.ClientPlugin.createPluginElement_());
Do we need to make sure the plugin is hidden?

[Jamie, 2014-08-12 22:35:58]

https://codereview.chromium.org/467903002/diff/1/remoting/webapp/client_plugi...
File remoting/webapp/client_plugin.js (right):

https://codereview.chromium.org/467903002/diff/1/remoting/webapp/client_plugi...
remoting/webapp/client_plugin.js:134:
document.body.appendChild(remoting.ClientPlugin.createPluginElement_());
On 2014/08/12 22:03:31, Wez wrote:
> Do we need to make sure the plugin is hidden?

It mustn't have the hidden attribute set, or Chrome won't load it. We create it
0x0 instead.

https://codereview.chromium.org/467903002/diff/1/remoting/webapp/client_plugi...
remoting/webapp/client_plugin.js:134:
document.body.appendChild(remoting.ClientPlugin.createPluginElement_());
We'll end up with 2 plugins; is that okay? It might be cleaner to set an onLoad
handler on the element (or some other mechanism of determining when it's ready).
The handler would delete the element, and we shouldn't create the real plugin
until the pre-load one is deleted. WDYT?

[Sergey Ulanov, 2014-08-15 01:00:18]

PTAL. added loadend event handler to remove the dummy plugin instance once it's
finished loading.

https://codereview.chromium.org/467903002/diff/1/remoting/webapp/client_plugi...
File remoting/webapp/client_plugin.js (right):

https://codereview.chromium.org/467903002/diff/1/remoting/webapp/client_plugi...
remoting/webapp/client_plugin.js:134:
document.body.appendChild(remoting.ClientPlugin.createPluginElement_());
On 2014/08/12 22:35:58, Jamie wrote:
> We'll end up with 2 plugins; is that okay?

Yes, the plugin handles multiple instances just fine.

> It might be cleaner to set an onLoad
> handler on the element (or some other mechanism of determining when it's
ready).
> The handler would delete the element, and we shouldn't create the real plugin
> until the pre-load one is deleted. WDYT?

It's not really necessary to block the creation of the real instance on the
preloading one - they should not interfere with each other. If user tries to
connect while chrome still compiles the plugin then chrome will wait until it's
compiled before it creates actual plugin instances (both instances will run in
the same process). I've added loadend even handled that deletes the plugin after
it's loaded. It doesn't look like it's worth keeping it around to use it later -
that would increase complexity here without much benefit (and we still need to
reset the plugin, e.g. for repeated connections).

[Jamie, 2014-08-15 19:00:07]

lgtm

[Sergey Ulanov, 2014-08-15 19:07:29]

The CQ bit was checked by sergeyu@chromium.org

[commit-bot: I haz the power, 2014-08-15 19:09:09]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/sergeyu@chromium.org/467903002/20001

[commit-bot: I haz the power, 2014-08-15 23:08:52]

Committed patchset #2 (20001) as 290040