Add android_ninja, like android_make for ninja.

Add android_ninja, like android_make for ninja.

android_ninja {tests,bench,gm,bench_pictures,render_pictures} all build and run
fine.  They build so fast I thought maybe they were broken.

Note that android_ninja with no argument is failing:
    ninja: Entering directory `out/config/android-arm_v7_thumb/Debug'
    ninja: error: 'lib.target/libSampleApp.so', needed by 'android/libs/armeabi-v7a/libSampleApp.so', missing and no known rule to make it
Will look into this.

BUG=
R=djsollen@google.com

Committed: https://code.google.com/p/skia/source/detail?r=11525

Committed: https://code.google.com/p/skia/source/detail?r=11527

[scroggo, 2013-09-26 21:23:24]

https://codereview.chromium.org/24833002/diff/5001/platform_tools/android/bin...
File platform_tools/android/bin/android_ninja (right):

https://codereview.chromium.org/24833002/diff/5001/platform_tools/android/bin...
platform_tools/android/bin/android_ninja:4: rm -f .android_config   # If no -d
flag is given, force the default device (not the last one).
Should all android building behave this way?

https://codereview.chromium.org/24833002/diff/5001/platform_tools/android/bin...
platform_tools/android/bin/android_ninja:8: exportVar GYP_GENERATORS ninja
Does this change my gyp generators for building non android?

[mtklein, 2013-09-26 21:39:54]

https://codereview.chromium.org/24833002/diff/5001/platform_tools/android/bin...
File platform_tools/android/bin/android_ninja (right):

https://codereview.chromium.org/24833002/diff/5001/platform_tools/android/bin...
platform_tools/android/bin/android_ninja:4: rm -f .android_config   # If no -d
flag is given, force the default device (not the last one).
On 2013/09/26 21:23:25, scroggo wrote:
> Should all android building behave this way?

Yep, this is only here because android_make does the same thing.

https://codereview.chromium.org/24833002/diff/5001/platform_tools/android/bin...
platform_tools/android/bin/android_ninja:8: exportVar GYP_GENERATORS ninja
On 2013/09/26 21:23:25, scroggo wrote:
> Does this change my gyp generators for building non android?

Nope.  In fact, android_make does the same thing setting it to make-android. 
I've refactored that a bit to make this clearer.

[scroggo, 2013-09-26 21:46:37]

On 2013/09/26 21:39:54, mtklein wrote:
>
https://codereview.chromium.org/24833002/diff/5001/platform_tools/android/bin...
> File platform_tools/android/bin/android_ninja (right):
> 
>
https://codereview.chromium.org/24833002/diff/5001/platform_tools/android/bin...
> platform_tools/android/bin/android_ninja:4: rm -f .android_config   # If no -d
> flag is given, force the default device (not the last one).
> On 2013/09/26 21:23:25, scroggo wrote:
> > Should all android building behave this way?
> 
> Yep, this is only here because android_make does the same thing.
> 
>
https://codereview.chromium.org/24833002/diff/5001/platform_tools/android/bin...
> platform_tools/android/bin/android_ninja:8: exportVar GYP_GENERATORS ninja
> On 2013/09/26 21:23:25, scroggo wrote:
> > Does this change my gyp generators for building non android?
> 
> Nope.  In fact, android_make does the same thing setting it to make-android. 
> I've refactored that a bit to make this clearer.

I still see a bad diff for android_make and android_setup.

[mtklein, 2013-09-27 13:55:55]

On 2013/09/26 21:46:37, scroggo wrote:
> On 2013/09/26 21:39:54, mtklein wrote:
> >
>
https://codereview.chromium.org/24833002/diff/5001/platform_tools/android/bin...
> > File platform_tools/android/bin/android_ninja (right):
> > 
> >
>
https://codereview.chromium.org/24833002/diff/5001/platform_tools/android/bin...
> > platform_tools/android/bin/android_ninja:4: rm -f .android_config   # If no
-d
> > flag is given, force the default device (not the last one).
> > On 2013/09/26 21:23:25, scroggo wrote:
> > > Should all android building behave this way?
> > 
> > Yep, this is only here because android_make does the same thing.
> > 
> >
>
https://codereview.chromium.org/24833002/diff/5001/platform_tools/android/bin...
> > platform_tools/android/bin/android_ninja:8: exportVar GYP_GENERATORS ninja
> > On 2013/09/26 21:23:25, scroggo wrote:
> > > Does this change my gyp generators for building non android?
> > 
> > Nope.  In fact, android_make does the same thing setting it to make-android.

> > I've refactored that a bit to make this clearer.
> 
> I still see a bad diff for android_make and android_setup.

Me too!  Try now?

[djsollen, 2013-09-27 13:57:38]

I can see them now.

[djsollen, 2013-09-27 14:07:50]

https://codereview.chromium.org/24833002/diff/35001/platform_tools/android/bi...
File platform_tools/android/bin/android_ninja (right):

https://codereview.chromium.org/24833002/diff/35001/platform_tools/android/bi...
platform_tools/android/bin/android_ninja:6: echo $DEVICE_ID > .android_config
seems like we could refactor android_make so that we could share not only this
logic, but the ccache logic as well.

https://codereview.chromium.org/24833002/diff/35001/platform_tools/android/bi...
platform_tools/android/bin/android_ninja:11: OUT=$SKIA_OUT/${BUILDTYPE-Debug}
The default string value is something I didn't know you could do, but it's just
what we need.  Can you just put a small comment there for us non-bash experts
that buildtype defaults to debug if not defined.

[mtklein, 2013-09-30 15:41:58]

https://codereview.chromium.org/24833002/diff/35001/platform_tools/android/bi...
File platform_tools/android/bin/android_ninja (right):

https://codereview.chromium.org/24833002/diff/35001/platform_tools/android/bi...
platform_tools/android/bin/android_ninja:6: echo $DEVICE_ID > .android_config
On 2013/09/27 14:07:50, djsollen wrote:
> seems like we could refactor android_make so that we could share not only this
> logic, but the ccache logic as well. 

Yup.  As we talked about before, all that ccache logic was dead code.  I've
refactored android_make to share as much logic as possible.  android_ninja is
just a symlink to android_make now.

https://codereview.chromium.org/24833002/diff/35001/platform_tools/android/bi...
platform_tools/android/bin/android_ninja:11: OUT=$SKIA_OUT/${BUILDTYPE-Debug}
On 2013/09/27 14:07:50, djsollen wrote:
> The default string value is something I didn't know you could do, but it's
just
> what we need.  Can you just put a small comment there for us non-bash experts
> that buildtype defaults to debug if not defined.

Yeah, it's handy.  Done.

[djsollen, 2013-09-30 15:50:32]

lgtm

[commit-bot: I haz the power, 2013-09-30 17:00:06]

CQ is trying da patch. Follow status at
https://skia-tree-status.appspot.com/cq/mtklein@google.com/24833002/42001

[commit-bot: I haz the power, 2013-09-30 17:10:06]

Failed to apply the patch.
svn: Commit failed (details follow):
svn: MKACTIVITY of '/svn/!svn/act/cadba964-0e45-4e71-9048-e2c565944b9c':
authorization failed: Could not authenticate to server: rejected Basic challenge
(https://skia.googlecode.com)

[mtklein, 2013-09-30 17:14:22]

Committed patchset #8 manually as r11525 (presubmit successful).

[mtklein, 2013-09-30 17:31:38]

On 2013/09/30 17:14:22, mtklein wrote:
> Committed patchset #8 manually as r11525 (presubmit successful).

Going to try again now having restored SCRIPT_DIR.

[mtklein, 2013-09-30 17:32:15]

Committed patchset #9 manually as r11527 (presubmit successful).