GN: Add system_wrappers_unittests, tools and tools_unittests

GN: Add system_wrappers_unittests, tools and tools_unittests

BUG=webrtc:5949
TESTED=Built and ran the tests on Mac.
NOTRY=True

Committed: https://crrev.com/bac0412e7f32ab8c7bfc14ddf263ccec20d92a37
Cr-Commit-Position: refs/heads/master@{#13007}

[kjellander_webrtc, 2016-06-01 15:53:50]

Description was changed from

==========
GN: Add system_wrappers_unittests, tools and tools_unittests

BUG=webrtc:5949
==========

to

==========
GN: Add system_wrappers_unittests, tools and tools_unittests

BUG=webrtc:5949
TESTED=Built and ran the tests on Mac.
==========

[kjellander_webrtc, 2016-06-01 15:53:50]

kjellander@webrtc.org changed reviewers:
+ phoglund@webrtc.org

[kjellander_webrtc, 2016-06-01 15:54:00]

Description was changed from

==========
GN: Add system_wrappers_unittests, tools and tools_unittests

BUG=webrtc:5949
TESTED=Built and ran the tests on Mac.
==========

to

==========
GN: Add system_wrappers_unittests, tools and tools_unittests

BUG=webrtc:5949
TESTED=Built and ran the tests on Mac.
NOTRY=True
==========

[kjellander_webrtc, 2016-06-01 17:21:04]

kjellander@webrtc.org changed reviewers:
+ henrika@webrtc.org

[kjellander_webrtc, 2016-06-01 17:21:04]

phoglund: main review.

henrika: webrtc/modules/audio_device/BUILD.gn - do you think that keypress code
is something that should be removed from audio_device_mac.cc? It seems wrong to
depend on CoreGraphics from audio-only code. My guess is that the library is
implicitly added for the GYP build, but not for GN, requiring it to be added
here (see compile error in PS#1).

[phoglund, 2016-06-02 07:51:45]

lgtm

[henrika_webrtc, 2016-06-02 07:52:16]

henrika@webrtc.org changed reviewers:
+ henrik.lundin@webrtc.org

[henrika_webrtc, 2016-06-02 07:56:22]

All platforms implement keypress detection and feed that state to the APM.

I don't know if it is used in the APM. If not, I guess it can be removed.
Adding hlundin@ to comment on the APM.

[hlundin-webrtc, 2016-06-02 09:12:28]

henrik.lundin@webrtc.org changed reviewers:
+ aluebs@webrtc.org
- henrik.lundin@webrtc.org

[hlundin-webrtc, 2016-06-02 09:12:28]

I don't know much about the keypress detection. I think aluebs@ knows more.

[kjellander_webrtc, 2016-06-02 09:17:08]

The CQ bit was checked by kjellander@webrtc.org

[kjellander_webrtc, 2016-06-02 09:17:09]

The patchset sent to the CQ was uploaded after l-g-t-m from phoglund@webrtc.org
Link to the patchset: https://codereview.webrtc.org/2025343002/#ps60001 (title:
"Rebased")

[commit-bot: I haz the power, 2016-06-02 09:17:15]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/2025343002/60001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/2025343002/60001

[kjellander_webrtc, 2016-06-02 09:17:53]

On 2016/06/02 09:12:28, hlundin-webrtc wrote:
> I don't know much about the keypress detection. I think aluebs@ knows more.

I'm committing this now, as fixing the keypress detection is a separate work
effort. I suggest you guys file a bug to track that in.

[commit-bot: I haz the power, 2016-06-02 09:18:53]

Description was changed from

==========
GN: Add system_wrappers_unittests, tools and tools_unittests

BUG=webrtc:5949
TESTED=Built and ran the tests on Mac.
NOTRY=True
==========

to

==========
GN: Add system_wrappers_unittests, tools and tools_unittests

BUG=webrtc:5949
TESTED=Built and ran the tests on Mac.
NOTRY=True
==========

[commit-bot: I haz the power, 2016-06-02 09:18:53]

Committed patchset #4 (id:60001)

[commit-bot: I haz the power, 2016-06-02 09:19:02]

Description was changed from

==========
GN: Add system_wrappers_unittests, tools and tools_unittests

BUG=webrtc:5949
TESTED=Built and ran the tests on Mac.
NOTRY=True
==========

to

==========
GN: Add system_wrappers_unittests, tools and tools_unittests

BUG=webrtc:5949
TESTED=Built and ran the tests on Mac.
NOTRY=True

Committed: https://crrev.com/bac0412e7f32ab8c7bfc14ddf263ccec20d92a37
Cr-Commit-Position: refs/heads/master@{#13007}
==========

[commit-bot: I haz the power, 2016-06-02 09:19:03]

Patchset 4 (id:??) landed as
https://crrev.com/bac0412e7f32ab8c7bfc14ddf263ccec20d92a37
Cr-Commit-Position: refs/heads/master@{#13007}

[henrika_webrtc, 2016-06-02 10:28:34]

kjellander@: not really sure what you mean by "fixing the keypress detection".
Your CL has not broken anything right? If the current implementation for Mac
blocks something, could you please file an issue for that (and assign to Alex)
since I don't know what the real problem is.

I don't think that any work will be done in the ADM for Mac to fix issues in
this area but
removing it for Mac is probably more probable (if needed).

[kjellander_webrtc, 2016-06-02 10:32:31]

On 2016/06/02 10:28:34, henrika_webrtc wrote:
> kjellander@: not really sure what you mean by "fixing the keypress detection".
> Your CL has not broken anything right? If the current implementation for Mac
> blocks something, could you please file an issue for that (and assign to Alex)
> since I don't know what the real problem is.
> 
> I don't think that any work will be done in the ADM for Mac to fix issues in
> this area but
> removing it for Mac is probably more probable (if needed).

I was thinking about this with the perspective of Slim and Modular design. We
shouldn't have audio code that depends on a graphics library, just because it
happens to use an API for detecting key press events that lives in the graphics
library.
It will add binary size and bloat to whoever wants to use audio but not video.

[henrika_webrtc, 2016-06-02 11:02:29]

AFAIK, CGEventSourceKeyState is required on Mac to detect global keypress
events.
I am convinced that alternative methods were investigated when it was first
added.

If Alex replies that keypress-detecion is not an essential part of the APM, we
can
most likely remove the support for all platforms. 
It was originally added to the ADM since a very fast detection was needed and
the audio
callback rate was a suitable rate. I don't think that this functionality is
essential
enough to account for any redesign at this stage.

FYI, Chrome uses CGEventSourceCounterForEventType() for this purpose and it
requires the
same dependency.

[henrika_webrtc, 2016-06-02 11:08:29]

Apple have bundled methods to detect low-level user input events in OS X
in the CoreGraphics framework and that is nothing we can can change. If keypress
detection is needed, the dependency is required.