Set is_test_apk=1 for content_browsertests_apk.

Set is_test_apk=1 for content_browsertests_apk.

Without this set, you won't get the android_tools target
built when building only content_browsertests_apk, which is needed
in order to run the test on a device.

In WebRTC, I'm adding buildbots using that target for building and
running content_browsertests in https://codereview.chromium.org/26738003/

I would like to avoid having to specify a separate target in
android_all.gyp just to get this build running, and I think all targets
should build their necessary dependencies.

TEST=local building of content_browsertests_apk + verifying the
host_forwarder was built.
BUG=305749
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=229632

[kjellander_chromium, 2013-10-17 12:26:53]

Hi, would this be OK to add?

[jochen (gone - plz use gerrit), 2013-10-17 22:39:06]

lgtm, but please wait for somebody from the android team to double check (maybe
bulach or beverloo)

[kjellander_chromium, 2013-10-18 08:46:07]

On 2013/10/17 22:39:06, jochen wrote:
> lgtm, but please wait for somebody from the android team to double check
(maybe
> bulach or beverloo)

OK, thanks.

bulach: can you also have a look?

[bulach, 2013-10-18 09:04:52]

lgtm, thanks!

just make sure the trybots and FYI bots will be happy, hopefully there were no
hidden dependency on this.

[commit-bot: I haz the power, 2013-10-18 19:01:35]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/kjellander@chromium.org/27622005/1

[commit-bot: I haz the power, 2013-10-18 20:23:40]

Step "update" is always a major failure.
Look at the try server FAQ for more details.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=linux_clan...

[commit-bot: I haz the power, 2013-10-19 19:17:58]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/kjellander@chromium.org/27622005/1

[commit-bot: I haz the power, 2013-10-19 22:35:20]

Change committed as 229632

[wjia(left Chromium), 2013-10-24 00:38:13]

This CL needs to be reverted since it has caused content_browsertests to fail to
run on android_fyi_dbg with the following message:
(http://build.chromium.org/p/tryserver.chromium/builders/android_fyi_dbg_trigg...)

I  212s Main  [F00A]> am start -a android.intent.action.MAIN -S -W -n
org.chromium.content_browsertests_apk/org.chromium.content_browsertests_apk.ContentBrowserTestsActivity
# 10/23/13 09:22:28: No response for adb -s 014E0FAC0500F00A shell 'am start -a
android.intent.action.MAIN -S -W -n
org.chromium.content_browsertests_apk/org.chromium.content_browsertests_apk.ContentBrowserTestsActivity',
retrying
# 10/23/13 09:22:48: No response for adb -s 014E0FAC0500F00A shell 'am start -a
android.intent.action.MAIN -S -W -n
org.chromium.content_browsertests_apk/org.chromium.content_browsertests_apk.ContentBrowserTestsActivity',
retrying
# 10/23/13 09:23:08: No response for adb -s 014E0FAC0500F00A shell 'am start -a
android.intent.action.MAIN -S -W -n
org.chromium.content_browsertests_apk/org.chromium.content_browsertests_apk.ContentBrowserTestsActivity',
retrying
W  292s Main  Failed obtaining tests from 014E0FAC0500F00A with exception: 
Traceback (most recent call last):
  File "build/android/test_runner.py", line 776, in <module>
    sys.exit(main(sys.argv))
  File "build/android/test_runner.py", line 772, in main
    return command_option_parser.ParseAndExecute(option_parser)
  File
"/b/build/slave/android/build/src/build/android/pylib/utils/command_option_parser.py",
line 75, in ParseAndExecute
    return cmd.run_command_func(argv[1], options, args, option_parser)
  File "build/android/test_runner.py", line 692, in RunTestsCommand
    return _RunGTests(options, option_parser.error, devices)
  File "build/android/test_runner.py", line 498, in _RunGTests
    runner_factory, tests = gtest_setup.Setup(gtest_options, devices)
  File "/b/build/slave/android/build/src/build/android/pylib/gtest/setup.py",
line 303, in Setup
    tests = _GetTestsFromDevice(TestRunnerFactory, devices)
  File "/b/build/slave/android/build/src/build/android/pylib/gtest/setup.py",
line 216, in _GetTestsFromDevice
    raise Exception('No device available to get the list of tests.')
Exception: No device available to get the list of tests.

[bulach, 2013-10-24 12:23:20]

hmm...

the build 2410 had the same error:
http://build.chromium.org/p/tryserver.chromium/builders/android_fyi_dbg_trigg...

it ran Fri Oct 18 15:50:58 2013, and the patch landed on Sat Oct 19 22:35:18...

The error message looks very unrelated to this patch:
Exception: No device available to get the list of tests.

it looks like the bot is having connectivity problems.

+frankf / navabi for further assessment

[wjia(left Chromium), 2013-10-24 16:39:42]

Marcus,

As you can see from your try job
http://build.chromium.org/p/tryserver.chromium/builders/android_fyi_dbg_trigg...,
content_browsertests are running now. But several of them are failing (6
are crashing and 2 timed out).

Your try job is using r230677 which includes my revert r230574.

Wei



On Thu, Oct 24, 2013 at 5:23 AM, <bulach@chromium.org> wrote:

> hmm...
>
> the build 2410 had the same error:
> http://build.chromium.org/p/**tryserver.chromium/builders/**
> android_fyi_dbg_triggered_**tests/builds/2410/steps/**
>
content_browsertests/logs/**stdio<http://build.chromium.org/p/tryserver.chromium/builders/android_fyi_dbg_triggered_tests/builds/2410/steps/content_browsertests/logs/stdio>
>
> it ran Fri Oct 18 15:50:58 2013, and the patch landed on Sat Oct 19
> 22:35:18...
>
> The error message looks very unrelated to this patch:
>
> Exception: No device available to get the list of tests.
>
> it looks like the bot is having connectivity problems.
>
> +frankf / navabi for further assessment
>
>
https://codereview.chromium.**org/27622005/<https://codereview.chromium.org/2...
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[bulach, 2013-10-24 17:16:36]

there are only 6 references to "is_test_apk" in the codebase:
https://code.google.com/p/chromium/codesearch#search/&q=is_test_apk&sq=packag...

I find it hard to believe (although not impossible!) that this change could have
any impact on raising Exception('No device available to get the list of tests.')

:)

frankf and navabi can probably provide more details, but I'd like to understand
why do you suspect this patch specifically? I'd be interested to learn how you
pinpoint this patch?

also, I don't quite understand what you meant by "my" try job... afaict, the try
job I linked, build 2410, had r229274, which was before this patch even landed,
no?


We do have lots of problems with our device (in)stability :) so I'm really
interested to learn how are you correlating this.

[wjia(left Chromium), 2013-10-24 17:25:02]

Please see comment #4: https://codereview.chromium.org/38753003/#msg4

By your try job, I mean build 2426 (
http://build.chromium.org/p/tryserver.chromium/builders/android_fyi_dbg_trigg...
).

I have bisected and found r229632 triggered the issue (content_browsertests
don't run). You can also try locally with r229632 and r229631. You need
change in clank.gyp added by fc2a682a6fa536e0bf9207ad47942524ed4d9a39 in
ssh://gerrit-int.chromium.org:29419/clank/internal/apps.git.



On Thu, Oct 24, 2013 at 10:16 AM, <bulach@chromium.org> wrote:

> there are only 6 references to "is_test_apk" in the codebase:
> https://code.google.com/p/**chromium/codesearch#search/&q=**
>
is_test_apk&sq=package:**chromium&type=cs<https://code.google.com/p/chromium/codesearch#search/&q=is_test_apk&sq=package:chromium&type=cs>
>
> I find it hard to believe (although not impossible!) that this change
> could have
> any impact on raising Exception('No device available to get the list of
> tests.')
>
> :)
>
> frankf and navabi can probably provide more details, but I'd like to
> understand
> why do you suspect this patch specifically? I'd be interested to learn how
> you
> pinpoint this patch?
>
> also, I don't quite understand what you meant by "my" try job... afaict,
> the try
> job I linked, build 2410, had r229274, which was before this patch even
> landed,
> no?
>
>
> We do have lots of problems with our device (in)stability :) so I'm really
> interested to learn how are you correlating this.
>
>
https://codereview.chromium.**org/27622005/<https://codereview.chromium.org/2...
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[bulach, 2013-10-24 17:40:27]

On 2013/10/24 17:25:02, wjia wrote:
> Please see comment #4: https://codereview.chromium.org/38753003/#msg4
> 
> By your try job, I mean build 2426 (
>
http://build.chromium.org/p/tryserver.chromium/builders/android_fyi_dbg_trigg...
> ).
> 
> I have bisected and found r229632 triggered the issue (content_browsertests
> don't run). 

I don't understand what's the correlation of build 2426?
I'm pointing out to build 2410 (r229274) which had the same error before this
patch landed...
Would you have a full logcat or anything else that could help debugging this
issue, and pinpoint why build r229274 was *different* from r229632? what am I
missing?


> 
> 
> 
> On Thu, Oct 24, 2013 at 10:16 AM, <mailto:bulach@chromium.org> wrote:
> 
> > there are only 6 references to "is_test_apk" in the codebase:
> > https://code.google.com/p/**chromium/codesearch#search/&q=**
> >
>
is_test_apk&sq=package:**chromium&type=cs<https://code.google.com/p/chromium/codesearch#search/&q=is_test_apk&sq=package:chromium&type=cs>
> >
> > I find it hard to believe (although not impossible!) that this change
> > could have
> > any impact on raising Exception('No device available to get the list of
> > tests.')
> >
> > :)
> >
> > frankf and navabi can probably provide more details, but I'd like to
> > understand
> > why do you suspect this patch specifically? I'd be interested to learn how
> > you
> > pinpoint this patch?
> >
> > also, I don't quite understand what you meant by "my" try job... afaict,
> > the try
> > job I linked, build 2410, had r229274, which was before this patch even
> > landed,
> > no?
> >
> >
> > We do have lots of problems with our device (in)stability :) so I'm really
> > interested to learn how are you correlating this.
> >
> >
>
https://codereview.chromium.**org/27622005/%3Chttps://codereview.chromium.org...>
> >
> 
> To unsubscribe from this group and stop receiving emails from it, send an
email
> to mailto:chromium-reviews+unsubscribe@chromium.org.

[wjia(left Chromium), 2013-10-24 17:51:44]

On Thu, Oct 24, 2013 at 10:40 AM, <bulach@chromium.org> wrote:

> On 2013/10/24 17:25:02, wjia wrote:
>
>> Please see comment #4:
https://codereview.chromium.**org/38753003/#msg4<https://codereview.chromium....
>>
>
>  By your try job, I mean build 2426 (
>>
>
> http://build.chromium.org/p/**tryserver.chromium/builders/**
>
android_fyi_dbg_triggered_**tests/builds/2426<http://build.chromium.org/p/tryserver.chromium/builders/android_fyi_dbg_triggered_tests/builds/2426>
>
>> ).
>>
>
>  I have bisected and found r229632 triggered the issue
>> (content_browsertests
>> don't run).
>>
>
> I don't understand what's the correlation of build 2426?
> I'm pointing out to build 2410 (r229274) which had the same error before
> this
> patch landed...
> Would you have a full logcat or anything else that could help debugging
> this
> issue, and pinpoint why build r229274 was *different* from r229632? what
> am I
> missing?
>
>
I am not sure what happened to build 2410.

My observation is that my local build with r229631 is able to run
content_browsertests,
while build with r229632 is not. This happens on the same device I am using
and multiple times.
I strongly recommend you try local build with r229631 and r229632, and
compare the difference.

I mentioned build 2426 because it includes r230574 which reverts r229632,
and it can run content_browsertests.

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[frankf, 2013-10-24 18:14:52]

AFAIK, is_test_apk is only used for instrumentation test apks. Comment in
java_apk.gyp:

is_test_apk - Set to 1 if building a test apk.  This prevents resources from
dependencies from being re-included.

You're using for another purpose though.

+cjhopman to verify.

[cjhopman, 2013-10-24 18:41:44]

On 2013/10/24 18:14:52, frankf wrote:
> AFAIK, is_test_apk is only used for instrumentation test apks. Comment in
> java_apk.gyp:
> 
> is_test_apk - Set to 1 if building a test apk.  This prevents resources from
> dependencies from being re-included.
> 
> You're using for another purpose though.
> 
> +cjhopman to verify.

Yes, I think setting is_test_apk has other side effects that you don't want.

While I agree that targets should list (at least) their actual dependencies,
it's not actually the case that content_browsertests_apk depends on
android_tools. You can build the apk, install it, and send intents to it to run
tests all without building android_tools. It is actually the user-friendly,
test-running scripts that require android_tools. That being said, I had added
the dependency on android_tools for instrumentation tests because I was annoyed
when I would build a test and then not be able to run it.

If setting is_test_apk is broken for non-instrumentation tests, there are a
couple of other options:

1. Add a new variable to condition the android_tools dependency
2. Add the android_tools dependency directly to the target (and not via the
.gypi)
3. Unconditionally depend on android_tools for all APKs
4. Explicitly build android_tools

I don't like (1). (4) is the current state, and it annoys me. I would probably
just say do (3). These targets change so infrequently and are so small that
building them when they aren't actually needed (i.e. when you don't want to run
tests) is not really a big cost (and I think it's a smaller cost than adding
more complexity to java_apk.gypi).

As for the test failures, I really, really don't think this could have caused
"Exception: No device available to get the list of tests.", though it's possible
that it would cause crashes when running the tests.

[frankf, 2013-10-24 18:53:59]

On 2013/10/24 18:41:44, cjhopman wrote:
> On 2013/10/24 18:14:52, frankf wrote:
> > AFAIK, is_test_apk is only used for instrumentation test apks. Comment in
> > java_apk.gyp:
> > 
> > is_test_apk - Set to 1 if building a test apk.  This prevents resources from
> > dependencies from being re-included.
> > 
> > You're using for another purpose though.
> > 
> > +cjhopman to verify.
> 
> Yes, I think setting is_test_apk has other side effects that you don't want.
> 
> While I agree that targets should list (at least) their actual dependencies,
> it's not actually the case that content_browsertests_apk depends on
> android_tools. You can build the apk, install it, and send intents to it to
run
> tests all without building android_tools. It is actually the user-friendly,
> test-running scripts that require android_tools. That being said, I had added
> the dependency on android_tools for instrumentation tests because I was
annoyed
> when I would build a test and then not be able to run it.
> 
> If setting is_test_apk is broken for non-instrumentation tests, there are a
> couple of other options:
> 
> 1. Add a new variable to condition the android_tools dependency
> 2. Add the android_tools dependency directly to the target (and not via the
> .gypi)
> 3. Unconditionally depend on android_tools for all APKs
> 4. Explicitly build android_tools
> 
> I don't like (1). (4) is the current state, and it annoys me. I would probably
> just say do (3). These targets change so infrequently and are so small that
> building them when they aren't actually needed (i.e. when you don't want to
run
> tests) is not really a big cost (and I think it's a smaller cost than adding
> more complexity to java_apk.gypi).
> 
> As for the test failures, I really, really don't think this could have caused
> "Exception: No device available to get the list of tests.", though it's
possible
> that it would cause crashes when running the tests.

Actually, that exception is misleading. It seems flipping that switch is causing
crashes related to emma coverage:

1800D:  10-23 23:55:58.898   219   456 I ActivityManager: Start proc
org.chromium.content_browsertests_apk for activity
org.chromium.content_browsertests_apk/.ContentBrowserTestsActivity: pid=21500
uid=10068 gids={1006, 3003, 1015}
1800D:  10-23 23:55:58.921 21500 21500 I dalvikvm: Could not find method
com.vladium.emma.rt.RT.r, referenced from method
org.chromium.content_browsertests_apk.ContentBrowserTestsApplication.$VRi
1800D:  10-23 23:55:58.921 21500 21500 W dalvikvm: VFY: unable to resolve static
method 1958: Lcom/vladium/emma/rt/RT;.r ([[ZLjava/lang/String;J)V
1800D:  10-23 23:55:58.921 21500 21500 D dalvikvm: VFY: replacing opcode 0x71 at
0x002b
1800D:  10-23 23:55:58.921 21500 21500 W dalvikvm: Exception
Ljava/lang/NoClassDefFoundError; thrown while initializing
Lorg/chromium/content_browsertests_apk/ContentBrowserTestsApplication;
1800D:  10-23 23:55:58.921 21500 21500 W dalvikvm: Class init failed in
newInstance call
(Lorg/chromium/content_browsertests_apk/ContentBrowserTestsApplication;)
1800D:  10-23 23:55:58.921 21500 21500 D AndroidRuntime: Shutting down VM
1800D:  10-23 23:55:58.921 21500 21500 W dalvikvm: threadid=1: thread exiting
with uncaught exception (group=0x40a631f8)
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: FATAL EXCEPTION: main
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime:
java.lang.ExceptionInInitializerError
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
java.lang.Class.newInstanceImpl(Native Method)
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
java.lang.Class.newInstance(Class.java:1319)
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
android.app.Instrumentation.newApplication(Instrumentation.java:957)
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
android.app.Instrumentation.newApplication(Instrumentation.java:942)
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
android.app.LoadedApk.makeApplication(LoadedApk.java:477)
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
android.app.ActivityThread.handleBindApplication(ActivityThread.java:3938)
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
android.app.ActivityThread.access$1300(ActivityThread.java:123)
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
android.app.ActivityThread$H.handleMessage(ActivityThread.java:1185)
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
android.os.Handler.dispatchMessage(Handler.java:99)
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
android.os.Looper.loop(Looper.java:137)
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
android.app.ActivityThread.main(ActivityThread.java:4424)
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
java.lang.reflect.Method.invokeNative(Native Method)
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
java.lang.reflect.Method.invoke(Method.java:511)
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
com.android.internal.os.ZygoteInit$MethodAndArgsCaller.run(ZygoteInit.java:784)
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
com.android.internal.os.ZygoteInit.main(ZygoteInit.java:551)
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
dalvik.system.NativeStart.main(Native Method)
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: Caused by:
java.lang.NoClassDefFoundError: com.vladium.emma.rt.RT
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
org.chromium.content_browsertests_apk.ContentBrowserTestsApplication.$VRi(ContentBrowserTestsApplication.java)
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
org.chromium.content_browsertests_apk.ContentBrowserTestsApplication.<clinit>(ContentBrowserTestsApplication.java)
1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	... 16 more
1800D:  10-23 23:55:58.929   219   219 W ActivityManager:   Force finishing
activity org.chromium.content_browsertests_apk/.ContentBrowserTestsActivity

[kjellander_chromium, 2013-10-25 13:35:54]

On 2013/10/24 18:53:59, frankf wrote:
> On 2013/10/24 18:41:44, cjhopman wrote:
> > On 2013/10/24 18:14:52, frankf wrote:
> > > AFAIK, is_test_apk is only used for instrumentation test apks. Comment in
> > > java_apk.gyp:
> > > 
> > > is_test_apk - Set to 1 if building a test apk.  This prevents resources
from
> > > dependencies from being re-included.
> > > 
> > > You're using for another purpose though.
> > > 
> > > +cjhopman to verify.
> > 
> > Yes, I think setting is_test_apk has other side effects that you don't want.
> > 
> > While I agree that targets should list (at least) their actual dependencies,
> > it's not actually the case that content_browsertests_apk depends on
> > android_tools. You can build the apk, install it, and send intents to it to
> run
> > tests all without building android_tools. It is actually the user-friendly,
> > test-running scripts that require android_tools. That being said, I had
added
> > the dependency on android_tools for instrumentation tests because I was
> annoyed
> > when I would build a test and then not be able to run it.
> > 
> > If setting is_test_apk is broken for non-instrumentation tests, there are a
> > couple of other options:
> > 
> > 1. Add a new variable to condition the android_tools dependency
> > 2. Add the android_tools dependency directly to the target (and not via the
> > .gypi)
> > 3. Unconditionally depend on android_tools for all APKs
> > 4. Explicitly build android_tools
> > 
> > I don't like (1). (4) is the current state, and it annoys me. I would
probably
> > just say do (3). These targets change so infrequently and are so small that
> > building them when they aren't actually needed (i.e. when you don't want to
> run
> > tests) is not really a big cost (and I think it's a smaller cost than adding
> > more complexity to java_apk.gypi).
> > 
> > As for the test failures, I really, really don't think this could have
caused
> > "Exception: No device available to get the list of tests.", though it's
> possible
> > that it would cause crashes when running the tests.
> 
> Actually, that exception is misleading. It seems flipping that switch is
causing
> crashes related to emma coverage:
> 
> 1800D:  10-23 23:55:58.898   219   456 I ActivityManager: Start proc
> org.chromium.content_browsertests_apk for activity
> org.chromium.content_browsertests_apk/.ContentBrowserTestsActivity: pid=21500
> uid=10068 gids={1006, 3003, 1015}
> 1800D:  10-23 23:55:58.921 21500 21500 I dalvikvm: Could not find method
> com.vladium.emma.rt.RT.r, referenced from method
> org.chromium.content_browsertests_apk.ContentBrowserTestsApplication.$VRi
> 1800D:  10-23 23:55:58.921 21500 21500 W dalvikvm: VFY: unable to resolve
static
> method 1958: Lcom/vladium/emma/rt/RT;.r ([[ZLjava/lang/String;J)V
> 1800D:  10-23 23:55:58.921 21500 21500 D dalvikvm: VFY: replacing opcode 0x71
at
> 0x002b
> 1800D:  10-23 23:55:58.921 21500 21500 W dalvikvm: Exception
> Ljava/lang/NoClassDefFoundError; thrown while initializing
> Lorg/chromium/content_browsertests_apk/ContentBrowserTestsApplication;
> 1800D:  10-23 23:55:58.921 21500 21500 W dalvikvm: Class init failed in
> newInstance call
> (Lorg/chromium/content_browsertests_apk/ContentBrowserTestsApplication;)
> 1800D:  10-23 23:55:58.921 21500 21500 D AndroidRuntime: Shutting down VM
> 1800D:  10-23 23:55:58.921 21500 21500 W dalvikvm: threadid=1: thread exiting
> with uncaught exception (group=0x40a631f8)
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: FATAL EXCEPTION: main
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime:
> java.lang.ExceptionInInitializerError
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
> java.lang.Class.newInstanceImpl(Native Method)
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
> java.lang.Class.newInstance(Class.java:1319)
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
> android.app.Instrumentation.newApplication(Instrumentation.java:957)
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
> android.app.Instrumentation.newApplication(Instrumentation.java:942)
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
> android.app.LoadedApk.makeApplication(LoadedApk.java:477)
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
> android.app.ActivityThread.handleBindApplication(ActivityThread.java:3938)
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
> android.app.ActivityThread.access$1300(ActivityThread.java:123)
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
> android.app.ActivityThread$H.handleMessage(ActivityThread.java:1185)
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
> android.os.Handler.dispatchMessage(Handler.java:99)
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
> android.os.Looper.loop(Looper.java:137)
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
> android.app.ActivityThread.main(ActivityThread.java:4424)
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
> java.lang.reflect.Method.invokeNative(Native Method)
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
> java.lang.reflect.Method.invoke(Method.java:511)
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
>
com.android.internal.os.ZygoteInit$MethodAndArgsCaller.run(ZygoteInit.java:784)
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
> com.android.internal.os.ZygoteInit.main(ZygoteInit.java:551)
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
> dalvik.system.NativeStart.main(Native Method)
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: Caused by:
> java.lang.NoClassDefFoundError: com.vladium.emma.rt.RT
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
>
org.chromium.content_browsertests_apk.ContentBrowserTestsApplication.$VRi(ContentBrowserTestsApplication.java)
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	at
>
org.chromium.content_browsertests_apk.ContentBrowserTestsApplication.<clinit>(ContentBrowserTestsApplication.java)
> 1800D:  10-23 23:55:58.921 21500 21500 E AndroidRuntime: 	... 16 more
> 1800D:  10-23 23:55:58.929   219   219 W ActivityManager:   Force finishing
> activity org.chromium.content_browsertests_apk/.ContentBrowserTestsActivity

Thanks all for putting in so much time into debugging this issue. 
I suggest we leave this reverted and add another helper build target instead to
use: https://codereview.chromium.org/43463004/
I'll leave it to you guys to sort out if further actions are desired to improve
the exceptions and/or clarify the effects of the is_test_apk variable, as I
don't think I can contribute so much to those efforts.