GN: Use a custom script for package_apk() build step rather than ANT

GN: Use a custom script for package_apk() build step rather than ANT

In addition to not having to rely on ANT, the script makes use of md5_check.py,
which detects that rebuilding *_incremental.apk as unnecessary in most cases.

BUG=375324, 533442
Committed: https://crrev.com/4716ea3a1b50d5071cc53c5a3b4ef652a58eab5f
Cr-Commit-Position: refs/heads/master@{#350746}

[agrieve, 2015-09-18 15:33:53]

agrieve@chromium.org changed reviewers:
+ yfriedman@chromium.org

[Yaron, 2015-09-24 17:25:18]

https://codereview.chromium.org/1348533007/diff/1/build/android/gyp/apkbuilde...
File build/android/gyp/apkbuilder.py (right):

https://codereview.chromium.org/1348533007/diff/1/build/android/gyp/apkbuilde...
build/android/gyp/apkbuilder.py:3: # Copyright (c) 2012 The Chromium Authors.
All rights reserved.
Can you share a design doc for your time machine?

https://codereview.chromium.org/1348533007/diff/1/build/android/gyp/apkbuilde...
build/android/gyp/apkbuilder.py:27: parser.add_argument('--dex-file',
It seems like this will be need to be a list in the case of multidex.

https://codereview.chromium.org/1348533007/diff/1/build/android/gyp/apkbuilde...
build/android/gyp/apkbuilder.py:59: tmp_apk = options.output_apk + '.tmp'
I guess this is guaranteed to be cleaned up if successful? If you fail to build
some apks, but not others, you would pollute the output location. Probably not
an issue in practice - I was just thinking about builder/tester split bots which
upload these unnecessarily. You could use a temp dir but I'll admit I don't feel
strongly about it.

https://codereview.chromium.org/1348533007/diff/1/build/android/gyp/apkbuilde...
build/android/gyp/apkbuilder.py:71: apk.write(options.dex_file, 'classes.dex')
How are multidex'd files named?

[agrieve, 2015-09-24 18:48:15]

https://codereview.chromium.org/1348533007/diff/1/build/android/gyp/apkbuilde...
File build/android/gyp/apkbuilder.py (right):

https://codereview.chromium.org/1348533007/diff/1/build/android/gyp/apkbuilde...
build/android/gyp/apkbuilder.py:3: # Copyright (c) 2012 The Chromium Authors.
All rights reserved.
On 2015/09/24 17:25:17, Yaron wrote:
> Can you share a design doc for your time machine?

Done.

https://codereview.chromium.org/1348533007/diff/1/build/android/gyp/apkbuilde...
build/android/gyp/apkbuilder.py:27: parser.add_argument('--dex-file',
On 2015/09/24 17:25:17, Yaron wrote:
> It seems like this will be need to be a list in the case of multidex.

multidex uses a single .zip file to pass multiple dex files. There is no
multidex logic in GN yet, so not easy to add logic for it yet. My plan for
follow-ups:

1. Add support for EMMA (crbug/535290)
2. Add support for multidex and hook it up to GYP (deleting ant.py)

https://codereview.chromium.org/1348533007/diff/1/build/android/gyp/apkbuilde...
build/android/gyp/apkbuilder.py:59: tmp_apk = options.output_apk + '.tmp'
On 2015/09/24 17:25:17, Yaron wrote:
> I guess this is guaranteed to be cleaned up if successful? If you fail to
build
> some apks, but not others, you would pollute the output location. Probably not
> an issue in practice - I was just thinking about builder/tester split bots
which
> upload these unnecessarily. You could use a temp dir but I'll admit I don't
feel
> strongly about it.

Changed it to use a proper temp file.

https://codereview.chromium.org/1348533007/diff/1/build/android/gyp/apkbuilde...
build/android/gyp/apkbuilder.py:71: apk.write(options.dex_file, 'classes.dex')
On 2015/09/24 17:25:17, Yaron wrote:
> How are multidex'd files named?

classes.dex, classes2.dex, classes3.dex...

[Yaron, 2015-09-24 18:56:11]

lgtm

https://codereview.chromium.org/1348533007/diff/1/build/android/gyp/apkbuilde...
File build/android/gyp/apkbuilder.py (right):

https://codereview.chromium.org/1348533007/diff/1/build/android/gyp/apkbuilde...
build/android/gyp/apkbuilder.py:59: tmp_apk = options.output_apk + '.tmp'
On 2015/09/24 18:48:15, agrieve wrote:
> On 2015/09/24 17:25:17, Yaron wrote:
> > I guess this is guaranteed to be cleaned up if successful? If you fail to
> build
> > some apks, but not others, you would pollute the output location. Probably
not
> > an issue in practice - I was just thinking about builder/tester split bots
> which
> > upload these unnecessarily. You could use a temp dir but I'll admit I don't
> feel
> > strongly about it.
> 
> Changed it to use a proper temp file.

There is tempfile.NamedTemporaryFile() which could've handled this with a "with"

[agrieve, 2015-09-24 19:09:19]

The CQ bit was checked by agrieve@chromium.org

[commit-bot: I haz the power, 2015-09-24 19:11:21]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1348533007/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1348533007/20001

[agrieve, 2015-09-24 19:47:02]

The CQ bit was unchecked by agrieve@chromium.org

[agrieve, 2015-09-25 02:10:29]

The CQ bit was checked by agrieve@chromium.org

[commit-bot: I haz the power, 2015-09-25 02:11:07]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1348533007/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1348533007/20001

[commit-bot: I haz the power, 2015-09-25 02:16:08]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-09-25 02:16:42]

Patchset 2 (id:??) landed as
https://crrev.com/4716ea3a1b50d5071cc53c5a3b4ef652a58eab5f
Cr-Commit-Position: refs/heads/master@{#350746}