Speech recognition: fire the speechstart/speechend events.

Speech recognition: fire the speechstart/speechend events.

I noticed that we never fire these events. To comply with the spec, we can simply fire them at the same time as soundstart/soundend.

BUG=116954

[hans, 2012-08-01 12:45:59]

satish: take a look?

[Satish, 2012-08-01 12:56:47]

I think speech start should be fired after the end pointer's environment
estimation has completed. Speech end can be fired together with sound end. WDYT?

[hans, 2012-08-01 13:14:32]

On 2012/08/01 12:56:47, Satish wrote:
> I think speech start should be fired after the end pointer's environment
> estimation has completed.

That's what this does. We fire sound start when the end pointer's estimation
thinks we're receiving speech ("endpointer_.DidStartReceivingSpeech()").

In our implementation, I don't think there's any way to tell the difference
between sound start and speech start, that's why I'm firing them together like
this.

[Satish, 2012-08-01 13:50:44]

lgtm

[hans, 2012-08-01 13:53:15]

jamesr: content/renderer/OWNERS review?

[jamesr, 2012-08-01 17:19:23]

By "spec" do you mean a web standard?  Why is this sort of logic in the chromium
repo?

[hans, 2012-08-01 17:21:43]

On 2012/08/01 17:19:23, jamesr wrote:
> By "spec" do you mean a web standard?  Why is this sort of logic in the
chromium
> repo?

Sorry, meant to have included a link. By "spec", I mean
http://dvcs.w3.org/hg/speech-api/raw-file/tip/speechapi.html#speechreco-events

[jamesr, 2012-08-01 17:42:38]

On 2012/08/01 17:21:43, hans wrote:
> On 2012/08/01 17:19:23, jamesr wrote:
> > By "spec" do you mean a web standard?  Why is this sort of logic in the
> chromium
> > repo?
> 
> Sorry, meant to have included a link. By "spec", I mean
> http://dvcs.w3.org/hg/speech-api/raw-file/tip/speechapi.html#speechreco-events

This call is not communicating any additional information from chromium to
WebKit.  Generally, I think code that has to do with satisfying a web standard
(as opposed to implementing the backend) should generally be in cross-platform
WebCore code if possible so that all implementations can be consistent.  It
seems like most of this code uses straight calls through to fire JS events.
Hmmmm...

[jamesr, 2012-08-01 17:43:15]

Darin - what's the right pattern here?  Synthesize the event here (in chromium)
or somewhere in WebCore?

[darin (slow to review), 2012-08-02 06:18:33]

I think you should just have one entry point for start and stop on the WebKit
API.  What is the use case for having two entry points per event?  Do you ever
need to call the methods independently?

https://chromiumcodereview.appspot.com/10825133/diff/1/content/renderer/speec...
File content/renderer/speech_recognition_dispatcher.cc (right):

https://chromiumcodereview.appspot.com/10825133/diff/1/content/renderer/speec...
content/renderer/speech_recognition_dispatcher.cc:113:
recognizer_client_->didStartSound(GetHandleFromID(request_id));
I don't understand why we need two function calls on recognizer_client_ here. 
What value does that add?

https://chromiumcodereview.appspot.com/10825133/diff/1/content/renderer/speec...
content/renderer/speech_recognition_dispatcher.cc:118:
recognizer_client_->didEndSpeech(GetHandleFromID(request_id));
Ditto.

[hans, 2012-08-02 11:01:23]

Sounds like it's better to do this on the WebKit side. I've moved it to
https://bugs.webkit.org/show_bug.cgi?id=92971

jamesr/darin: please take a look.

[darin (slow to review), 2012-08-02 16:48:43]

Great, thanks!

On Thu, Aug 2, 2012 at 4:01 AM, <hans@chromium.org> wrote:

> Sounds like it's better to do this on the WebKit side. I've moved it to
>
https://bugs.webkit.org/show_**bug.cgi?id=92971<https://bugs.webkit.org/show_...
>
> jamesr/darin: please take a look.
>
>
https://chromiumcodereview.**appspot.com/10825133/<https://chromiumcodereview...
>

[Satish, 2012-09-02 22:24:06]

still need this CL?

[hans, 2012-09-03 12:33:16]

On 2012/09/02 22:24:06, Satish wrote:
> still need this CL?

Nope, it was fixed in WebKit.

Closing.