Add monkey command to android_commands.

Add monkey command to android_commands.
    
Add the option of reinstalling using -r instead of
uinstalling then installing.

BUG=140947
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=153093

[frankf, 2012-08-10 23:59:38]

@Marcus, what's the best way to fjord the monkey tests on the available devices
with maximum code reuse? Currently most of the test runner/sharders are tied to
the type of test (i.e. instrumentation or unit_tests).

Is there a better way than duplicating parts of run_python_tests.py but passing
it a PythonTestBase class with a single test method? 

Also, Trade Federation also has a monkey test runner, but since this a pretty
standard adb shell command, I think it belongs here.

[frankf, 2012-08-21 21:56:46]

@ilevy, please take a look. Is ManagedInstall called anywhere?

[Isaac (away), 2012-08-22 02:03:28]

Yes it is called:

http://code.google.com/p/chromium/source/search?q=ManagedInstall&btnG=Search+...

http://codereview.chromium.org/10829289/diff/4001/build/android/pylib/android...
File build/android/pylib/android_commands.py (right):

http://codereview.chromium.org/10829289/diff/4001/build/android/pylib/android...
build/android/pylib/android_commands.py:338: reinstall: Whether to reinstall
instead of uninstalling first.
I know I said it was OK earlier but I think reinstall is more confusing than
keep_data.  Would rather keep the current param name and add more info in the
comment here about what it does (i.e. if false, uninstall/install, if true,
install -r.

http://codereview.chromium.org/10829289/diff/4001/build/android/pylib/android...
build/android/pylib/android_commands.py:1011: if category is None:
the preferred way of doing this is:

category = category or []

http://codereview.chromium.org/10829289/diff/4001/build/android/pylib/android...
build/android/pylib/android_commands.py:1015: seed = random.randint(1, 100)
seed = seed or random.randint(1, 100)

http://codereview.chromium.org/10829289/diff/4001/build/android/pylib/android...
build/android/pylib/android_commands.py:1022: ' '.join(['-v'] * verbose),
don't like this syntax -- can you figure out an alternate way to pass an
optional '-v' ?

[frankf, 2012-08-22 21:46:54]

PTAL.

http://codereview.chromium.org/10829289/diff/4001/build/android/pylib/android...
File build/android/pylib/android_commands.py (right):

http://codereview.chromium.org/10829289/diff/4001/build/android/pylib/android...
build/android/pylib/android_commands.py:338: reinstall: Whether to reinstall
instead of uninstalling first.
On 2012/08/22 02:03:28, Isaac wrote:
> I know I said it was OK earlier but I think reinstall is more confusing than
> keep_data.  Would rather keep the current param name and add more info in the
> comment here about what it does (i.e. if false, uninstall/install, if true,
> install -r.

Done.

http://codereview.chromium.org/10829289/diff/4001/build/android/pylib/android...
build/android/pylib/android_commands.py:1011: if category is None:
On 2012/08/22 02:03:28, Isaac wrote:
> the preferred way of doing this is:
> 
> category = category or []

Done.

http://codereview.chromium.org/10829289/diff/4001/build/android/pylib/android...
build/android/pylib/android_commands.py:1015: seed = random.randint(1, 100)
On 2012/08/22 02:03:28, Isaac wrote:
> seed = seed or random.randint(1, 100)

Done.

http://codereview.chromium.org/10829289/diff/4001/build/android/pylib/android...
build/android/pylib/android_commands.py:1022: ' '.join(['-v'] * verbose),
The only other way is restrict verbose to a bool (i.e. either nothing or -v -v
-v). This is more flexible. Renamed verbose->verbosity to make it clear it's not
a bool.

On 2012/08/22 02:03:28, Isaac wrote:
> don't like this syntax -- can you figure out an alternate way to pass an
> optional '-v' ?

[Isaac (away), 2012-08-22 21:53:18]

http://codereview.chromium.org/10829289/diff/4001/build/android/pylib/android...
File build/android/pylib/android_commands.py (right):

http://codereview.chromium.org/10829289/diff/4001/build/android/pylib/android...
build/android/pylib/android_commands.py:1022: ' '.join(['-v'] * verbose),
On 2012/08/22 21:46:55, frankf wrote:
> The only other way is restrict verbose to a bool (i.e. either nothing or -v -v
> -v). This is more flexible. Renamed verbose->verbosity to make it clear it's
not
> a bool.
> 
> On 2012/08/22 02:03:28, Isaac wrote:
> > don't like this syntax -- can you figure out an alternate way to pass an
> > optional '-v' ?
> 

Oh, I didn't realize you wanted to pass multiple -v.

How about ' -v' * verbosity

[frankf, 2012-08-22 22:05:44]

Ah, one more reason Python's awesome.

[Isaac (away), 2012-08-23 02:30:07]

+cc cmp

lgtm w/ nit.  needs lgtm from full committer.

https://chromiumcodereview.appspot.com/10829289/diff/14002/build/android/pyli...
File build/android/pylib/android_commands.py (right):

https://chromiumcodereview.appspot.com/10829289/diff/14002/build/android/pyli...
build/android/pylib/android_commands.py:329: install_cmd = ' '.join(install_cmd)
nit: please use fresh variable here

[cmp, 2012-08-23 05:29:29]

lgtm

[commit-bot: I haz the power, 2012-08-23 19:14:36]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/frankf@chromium.org/10829289/14002

[commit-bot: I haz the power, 2012-08-23 22:32:39]

Change committed as 153093

[bulach, 2012-08-28 11:08:34]

lgtm

sorry about the delay, I was OOO for a bit..

as for sharding / fjording: it is indeed reasonably tied to the type of test:
gtests we simply split the test list into N shards, instrumentation we create a
queue and then each shard will fetch from it, webkit has its own mechanism...

I'm not sure how deterministic are these monkey tests, specially in terms of how
long it takes to set up and execute: if the amount of time is somewhat constant,
than it's simpler to just split the list and run them...

if it's variable, or if the setup for each test is significant, then it's
probably best to have a queue..

not sure if this answers your question :) but I'm happy to give more details on
all complications we had in the past..