Add dep for playground

Add dep for playground

[zhurunz, 2010-03-24 18:58:15]

1. More playground to dep, since will have more tests based on that.
2. Several improvements.

[petkov, 2010-03-24 19:17:22]

http://codereview.chromium.org/1224003/diff/12001/8008
File client/deps/realtimecomm_playground/pgutil.py (right):

http://codereview.chromium.org/1224003/diff/12001/8008#newcode1
client/deps/realtimecomm_playground/pgutil.py:1: # Copyright (c) 2010 The
Chromium OS Authors. All rights reserved.
Again, you should consider adding this to utils and upstreaming (i.e., sending a
patch to mbligh cc autotest@test.kernel.org).

http://codereview.chromium.org/1224003/diff/12001/8008#newcode17
client/deps/realtimecomm_playground/pgutil.py:17: def
get_number_of_logical_cpu():
Two empty lines between routines.

http://codereview.chromium.org/1224003/diff/12001/8010
File
client/site_tests/realtimecomm_GTalkPlayground/realtimecomm_GTalkPlayground.py
(right):

http://codereview.chromium.org/1224003/diff/12001/8010#newcode19
client/site_tests/realtimecomm_GTalkPlayground/realtimecomm_GTalkPlayground.py:19:
def setup(self):
Hmm, it seems that your new setup violates the convention we have for setups --
setup methods run on the build host, not on the client. Is my understanding
about your setup correct?

http://codereview.chromium.org/1224003/diff/12001/8010#newcode103
client/site_tests/realtimecomm_GTalkPlayground/realtimecomm_GTalkPlayground.py:103:
import pgutil
Any reason for importing here?

[zhurunz, 2010-03-24 19:38:50]

Fixes + comments.

http://codereview.chromium.org/1224003/diff/12001/8008
File client/deps/realtimecomm_playground/pgutil.py (right):

http://codereview.chromium.org/1224003/diff/12001/8008#newcode1
client/deps/realtimecomm_playground/pgutil.py:1: # Copyright (c) 2010 The
Chromium OS Authors. All rights reserved.
On 2010/03/24 19:17:22, petkov wrote:
> Again, you should consider adding this to utils and upstreaming (i.e., sending
a
> patch to mbligh cc mailto:autotest@test.kernel.org).

OK. Let me move those things in a separate CL.

http://codereview.chromium.org/1224003/diff/12001/8008#newcode17
client/deps/realtimecomm_playground/pgutil.py:17: def
get_number_of_logical_cpu():
On 2010/03/24 19:17:22, petkov wrote:
> Two empty lines between routines.
> 

Done.

http://codereview.chromium.org/1224003/diff/12001/8010
File
client/site_tests/realtimecomm_GTalkPlayground/realtimecomm_GTalkPlayground.py
(right):

http://codereview.chromium.org/1224003/diff/12001/8010#newcode19
client/site_tests/realtimecomm_GTalkPlayground/realtimecomm_GTalkPlayground.py:19:
def setup(self):
On 2010/03/24 19:17:22, petkov wrote:
> Hmm, it seems that your new setup violates the convention we have for setups
--
> setup methods run on the build host, not on the client. Is my understanding
> about your setup correct?
> 
I was doing the similar thing in glbench. 
My setup simply untar a file. I though it was done on host and then copied to
target machine. (Correct me if my understanding it wrong.)

http://codereview.chromium.org/1224003/diff/12001/8010#newcode103
client/site_tests/realtimecomm_GTalkPlayground/realtimecomm_GTalkPlayground.py:103:
import pgutil
On 2010/03/24 19:17:22, petkov wrote:
> Any reason for importing here?
> 

I tried moving it to the beginning and not successful, since it needs the
self.autodir here.

[petkov, 2010-03-24 21:42:38]

On 2010/03/24 19:38:50, zhurunz wrote:
> Fixes + comments.
> 
> http://codereview.chromium.org/1224003/diff/12001/8008
> File client/deps/realtimecomm_playground/pgutil.py (right):
> 
> http://codereview.chromium.org/1224003/diff/12001/8008#newcode1
> client/deps/realtimecomm_playground/pgutil.py:1: # Copyright (c) 2010 The
> Chromium OS Authors. All rights reserved.
> On 2010/03/24 19:17:22, petkov wrote:
> > Again, you should consider adding this to utils and upstreaming (i.e.,
sending
> a
> > patch to mbligh cc mailto:autotest@test.kernel.org).
> 
> OK. Let me move those things in a separate CL.
> 
> http://codereview.chromium.org/1224003/diff/12001/8008#newcode17
> client/deps/realtimecomm_playground/pgutil.py:17: def
> get_number_of_logical_cpu():
> On 2010/03/24 19:17:22, petkov wrote:
> > Two empty lines between routines.
> > 
> 
> Done.
> 
> http://codereview.chromium.org/1224003/diff/12001/8010
> File
> client/site_tests/realtimecomm_GTalkPlayground/realtimecomm_GTalkPlayground.py
> (right):
> 
> http://codereview.chromium.org/1224003/diff/12001/8010#newcode19
>
client/site_tests/realtimecomm_GTalkPlayground/realtimecomm_GTalkPlayground.py:19:
> def setup(self):
> On 2010/03/24 19:17:22, petkov wrote:
> > Hmm, it seems that your new setup violates the convention we have for setups
> --
> > setup methods run on the build host, not on the client. Is my understanding
> > about your setup correct?
> > 
> I was doing the similar thing in glbench. 
> My setup simply untar a file. I though it was done on host and then copied to
> target machine. (Correct me if my understanding it wrong.)

glbench deps has a setup method that runs on the host. Your deps doesn't have a
setup method. Then as part of run_once, gl_Bench installs the deps on the
client.

I'm not sure what happens in your case, does your test work? Do you happen to
get /home/chronos/ under your chroot (you shouldn't)?

> 
> http://codereview.chromium.org/1224003/diff/12001/8010#newcode103
>
client/site_tests/realtimecomm_GTalkPlayground/realtimecomm_GTalkPlayground.py:103:
> import pgutil
> On 2010/03/24 19:17:22, petkov wrote:
> > Any reason for importing here?
> > 
> 
> I tried moving it to the beginning and not successful, since it needs the
> self.autodir here.

[zhurunz, 2010-03-25 01:12:38]

Please take another review.

[zhurunz1, 2010-03-25 19:22:28]

I have moved "untar" to build machine.
The client just do a copytree now.
Please take another review.

[petkov, 2010-03-25 19:28:19]

lgtm

http://codereview.chromium.org/1224003/diff/27001/28005
File
client/site_tests/realtimecomm_GTalkPlayground/realtimecomm_GTalkPlayground.py
(right):

http://codereview.chromium.org/1224003/diff/27001/28005#newcode34
client/site_tests/realtimecomm_GTalkPlayground/realtimecomm_GTalkPlayground.py:34:
def run_setup(self, tarball='GTalkPlayground.tar.gz'):
Remove the tarball argument -- it's not used any more.

http://codereview.chromium.org/1224003/diff/27001/28005#newcode36
client/site_tests/realtimecomm_GTalkPlayground/realtimecomm_GTalkPlayground.py:36:
utils.system('rm -rf %s' % self.playground)
You could use shutil.rmtree here.