End-to-end browser tests for speech recognition.

End-to-end browser tests for speech recognition. (Speech CL2.9)

- Performing end-to-end browser tests, currently only on one-shot speech recognition, putting all the speech recognition infrastructure together and mocking at the very far endpoints (audio controller and URL fetcher).
- Moved speech_recognition_browsertest.cc from /content to /chrome due to its dependency with speech_recognition_bubble_controller.cc

BUG=116954
TEST=SpeechRecognitionBrowserTest

[Primiano Tucci (use gerrit), 2012-07-11 15:46:08]

Note: this CL depends on (is diff-ed from):
http://codereview.chromium.org/10702141 and
http://codereview.chromium.org/10704154

[jam, 2012-07-11 15:58:59]

> - Moved speech_recognition_browsertest.cc from /content to /chrome due to its
dependency with speech_recognition_bubble_controller.cc 

content should have tests that live in content (i.e. just like we don't want to
test base module inside of chrome), so moving this test from content to chrome
is something that's to be avoided.

if there are chrome specific tests for this area, i.e. those that want to ensure
that the chrome UI works as expected, it's fine that those are in chrome.

[jam, 2012-07-11 15:59:04]

https://chromiumcodereview.appspot.com/10703141/diff/2001/chrome/browser/spee...
File chrome/browser/speech/speech_recognition_bubble_controller.h (right):

https://chromiumcodereview.appspot.com/10703141/diff/2001/chrome/browser/spee...
chrome/browser/speech/speech_recognition_bubble_controller.h:39: class
DelegateForTests {
I haven't seen this pattern before. it seems unfortunate to add this interface
and code just for tests. is there no other way to get at this information?

[Primiano Tucci (use gerrit), 2012-07-11 16:15:34]

On 2012/07/11 15:58:59, John Abd-El-Malek wrote:
> > - Moved speech_recognition_browsertest.cc from /content to /chrome due to
its
> dependency with speech_recognition_bubble_controller.cc 
> 
> content should have tests that live in content (i.e. just like we don't want
to
> test base module inside of chrome), so moving this test from content to chrome
> is something that's to be avoided.
> 
> if there are chrome specific tests for this area, i.e. those that want to
ensure
> that the chrome UI works as expected, it's fine that those are in chrome.

Well, the only file moved to /chrome is speech_recognition_browsertests. All its
test fixtures are end-to-end tests which check the correctness of the speech
recognition by looking at the evolution of the bubble, thus why it has been
moved to /chrome.
Is it ok to do that under this circumstance?

[Primiano Tucci (use gerrit), 2012-07-11 16:15:49]

https://chromiumcodereview.appspot.com/10703141/diff/2001/chrome/browser/spee...
File chrome/browser/speech/speech_recognition_bubble_controller.h (right):

https://chromiumcodereview.appspot.com/10703141/diff/2001/chrome/browser/spee...
chrome/browser/speech/speech_recognition_bubble_controller.h:39: class
DelegateForTests {
On 2012/07/11 15:59:04, John Abd-El-Malek wrote:
> I haven't seen this pattern before. it seems unfortunate to add this interface
> and code just for tests. is there no other way to get at this information?
Unfortunately, I have no other idea on how to check the UI behavior of the
bubble other than intercepting events from it :/
Any clues?

[jam, 2012-07-11 16:33:24]

On 2012/07/11 16:15:34, Primiano Tucci wrote:
> On 2012/07/11 15:58:59, John Abd-El-Malek wrote:
> > > - Moved speech_recognition_browsertest.cc from /content to /chrome due to
> its
> > dependency with speech_recognition_bubble_controller.cc 
> > 
> > content should have tests that live in content (i.e. just like we don't want
> to
> > test base module inside of chrome), so moving this test from content to
chrome
> > is something that's to be avoided.
> > 
> > if there are chrome specific tests for this area, i.e. those that want to
> ensure
> > that the chrome UI works as expected, it's fine that those are in chrome.
> 
> Well, the only file moved to /chrome is speech_recognition_browsertests. All
its
> test fixtures are end-to-end tests which check the correctness of the speech
> recognition by looking at the evolution of the bubble, thus why it has been
> moved to /chrome.


That file that's currently in content has 3 tests, and those currently don't
depend on the speech bubble. Those tests are being deleted, with other tests now
in chrome that test stuff using the bubble.

Am I missing something?

> Is it ok to do that under this circumstance?

[Primiano Tucci (use gerrit), 2012-07-11 16:46:04]

On 2012/07/11 16:33:24, John Abd-El-Malek wrote:
> On 2012/07/11 16:15:34, Primiano Tucci wrote:
> > On 2012/07/11 15:58:59, John Abd-El-Malek wrote:
> > > > - Moved speech_recognition_browsertest.cc from /content to /chrome due
to
> > its
> > > dependency with speech_recognition_bubble_controller.cc 
> > > 
> > > content should have tests that live in content (i.e. just like we don't
want
> > to
> > > test base module inside of chrome), so moving this test from content to
> chrome
> > > is something that's to be avoided.
> > > 
> > > if there are chrome specific tests for this area, i.e. those that want to
> > ensure
> > > that the chrome UI works as expected, it's fine that those are in chrome.
> > 
> > Well, the only file moved to /chrome is speech_recognition_browsertests. All
> its
> > test fixtures are end-to-end tests which check the correctness of the speech
> > recognition by looking at the evolution of the bubble, thus why it has been
> > moved to /chrome.
> 
> 
> That file that's currently in content has 3 tests, and those currently don't
> depend on the speech bubble. Those tests are being deleted, with other tests
now
> in chrome that test stuff using the bubble.
> 
> Am I missing something?

No, it's correct.