Adding support for launching firefox from webrtc.DEPS.

Adding support for launching firefox from webrtc.DEPS.

This is required for the firefox interop test being added in
https://codereview.chromium.org/111383003/.

This is only intended to work for Linux for now. We will need some light
redesign to get this to work on Windows as well.

R=kjellander@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=240895

[kjellander_chromium, 2013-12-12 18:27:47]

drive-by before you publish :)
I think you should put the manifestdestiny and moz* packages directly in
svn:/​/​svn.chromium.org/​chrome/​trunk/​deps/​third_party/ instead of hiding it
below the webrtc/webrtc.DEPS folder. That way others could see and use it more
easily.

[phoglund_chromium, 2013-12-13 15:17:55]

Please review. Will make a separate patch where I check in the moz libraries in
deps/.

[kjellander_chromium, 2013-12-16 12:59:04]

https://codereview.chromium.org/114293002/diff/80001/DEPS
File DEPS (right):

https://codereview.chromium.org/114293002/diff/80001/DEPS#newcode14
DEPS:14: "/trunk/deps/third_party/mozdownload",
I suggest you setup pinned revisions for these as soon you've checked them into
deps/

https://codereview.chromium.org/114293002/diff/80001/DEPS#newcode17
DEPS:17: "webrtc.DEPS/third_party/manifestdestiny":
put manifestdestiny above the moz* ones to preserve a bit of alphabetical
sorting.

https://codereview.chromium.org/114293002/diff/80001/download_firefox_nightly.py
File download_firefox_nightly.py (right):

https://codereview.chromium.org/114293002/diff/80001/download_firefox_nightly...
download_firefox_nightly.py:2: #  Copyright (c) 2013 The WebRTC project authors.
All Rights Reserved.
Please change to Chromium's license.

https://codereview.chromium.org/114293002/diff/80001/download_firefox_nightly...
download_firefox_nightly.py:28: 
nit: two blank lines before top-level function.

https://codereview.chromium.org/114293002/diff/80001/run_firefox_webrtc.py
File run_firefox_webrtc.py (right):

https://codereview.chromium.org/114293002/diff/80001/run_firefox_webrtc.py#ne...
run_firefox_webrtc.py:2: #  Copyright (c) 2013 The WebRTC project authors. All
Rights Reserved.
Please change to Chromium's license.

https://codereview.chromium.org/114293002/diff/80001/run_firefox_webrtc.py#ne...
run_firefox_webrtc.py:32: 'media.navigator.enabled': True,
You might want to try without enabling 'media.navigator.enabled' and
'media.peerconnection.enabled' since I suspect at least the former is enabled by
default by now.

https://codereview.chromium.org/114293002/diff/80001/run_firefox_webrtc.py#ne...
run_firefox_webrtc.py:63: firefox_runner.process_handler.wait(timeout=60)
How do we motivate 60 seconds here? I assume we might have longer-running tests
in the future. Maybe making this setting a script flag with a default value of
60 could reduce the risk of getting confused about suddenly killed Firefox
processes in the future?

[phoglund_chromium, 2013-12-16 14:38:45]

https://codereview.chromium.org/114293002/diff/80001/DEPS
File DEPS (right):

https://codereview.chromium.org/114293002/diff/80001/DEPS#newcode14
DEPS:14: "/trunk/deps/third_party/mozdownload",
On 2013/12/16 12:59:04, Henrik Kjellander wrote:
> I suggest you setup pinned revisions for these as soon you've checked them
into
> deps/
Ok, I've sent you that patch for review. Let's get that in first.

https://codereview.chromium.org/114293002/diff/80001/DEPS#newcode17
DEPS:17: "webrtc.DEPS/third_party/manifestdestiny":
On 2013/12/16 12:59:04, Henrik Kjellander wrote:
> put manifestdestiny above the moz* ones to preserve a bit of alphabetical
> sorting.

Done.

https://codereview.chromium.org/114293002/diff/80001/download_firefox_nightly.py
File download_firefox_nightly.py (right):

https://codereview.chromium.org/114293002/diff/80001/download_firefox_nightly...
download_firefox_nightly.py:2: #  Copyright (c) 2013 The WebRTC project authors.
All Rights Reserved.
On 2013/12/16 12:59:04, Henrik Kjellander wrote:
> Please change to Chromium's license. 

Done.

https://codereview.chromium.org/114293002/diff/80001/download_firefox_nightly...
download_firefox_nightly.py:28: 
On 2013/12/16 12:59:04, Henrik Kjellander wrote:
> nit: two blank lines before top-level function.

Done.

https://codereview.chromium.org/114293002/diff/80001/run_firefox_webrtc.py
File run_firefox_webrtc.py (right):

https://codereview.chromium.org/114293002/diff/80001/run_firefox_webrtc.py#ne...
run_firefox_webrtc.py:2: #  Copyright (c) 2013 The WebRTC project authors. All
Rights Reserved.
On 2013/12/16 12:59:04, Henrik Kjellander wrote:
> Please change to Chromium's license. 

Done.

https://codereview.chromium.org/114293002/diff/80001/run_firefox_webrtc.py#ne...
run_firefox_webrtc.py:32: 'media.navigator.enabled': True,
On 2013/12/16 12:59:04, Henrik Kjellander wrote:
> You might want to try without enabling 'media.navigator.enabled' and
> 'media.peerconnection.enabled' since I suspect at least the former is enabled
by
> default by now.

Done.

https://codereview.chromium.org/114293002/diff/80001/run_firefox_webrtc.py#ne...
run_firefox_webrtc.py:63: firefox_runner.process_handler.wait(timeout=60)
On 2013/12/16 12:59:04, Henrik Kjellander wrote:
> How do we motivate 60 seconds here? I assume we might have longer-running
tests
> in the future. Maybe making this setting a script flag with a default value of
> 60 could reduce the risk of getting confused about suddenly killed Firefox
> processes in the future?
I'm going to have to redesign this thing for Windows anyway, so I suggest we
leave this as-is for now. The new design will probably have some mechanism to
look for a successful call on the Firefox side, which renders this timeout
uninteresting. For now, I actually think we should increase the timeout to
something very high and thereby make it chrome's decision to kill FF (which is
how the test is designed today).

[kjellander_chromium, 2013-12-16 14:45:38]

lgtm assuming you change to the DEPS-pinned revisions for the Mozilla
dependencies before committing.

[phoglund_chromium, 2013-12-16 15:11:57]

Pinned revisions + tested that it still works. Added mozprofile which I forgot.

[phoglund_chromium, 2013-12-16 15:12:20]

Committed patchset #8 manually as r240895 (presubmit successful).