Take the juju API address from the environment

Take the juju API address from the environment

The charm now tries to retrieve the juju API
address from the hook context before parsing
the machiner agent file.

This way we take advantage of a change in juju-core 
currently in review, which introduces the corresponding
functionality.

https://code.launchpad.net/~frankban/charms/precise/juju-gui/api-address/+merge/166808

(do not edit description out of merge proposal)

[frankban, 2013-05-31 14:29:10]

Please take a look.

[teknico, 2013-05-31 15:22:04]

LGTM, nice code and tests. A few trivials.

https://codereview.appspot.com/9810044/diff/1/hooks/utils.py
File hooks/utils.py (right):

https://codereview.appspot.com/9810044/diff/1/hooks/utils.py#newcode135
hooks/utils.py:135: base_dir = os.path.abspath(base_dir)
Sorry, can't resist: :-)

    segs = (CURRENT_DIR, '..', '..') if unit_dir is None else (unit_dir, '..')
    base_dir = os.path.abspath(os.path.join(*segs))

https://codereview.appspot.com/9810044/diff/1/tests/test_utils.py
File tests/test_utils.py (right):

https://codereview.appspot.com/9810044/diff/1/tests/test_utils.py#newcode97
tests/test_utils.py:97: If addresses are provided, also create a machiner
directory and an
What's a "machiner"?

https://codereview.appspot.com/9810044/diff/1/tests/test_utils.py#newcode133
tests/test_utils.py:133: # The API address is correctly retrieved from the
machiner agent file.
"machiner" again.

[bac, 2013-06-03 13:08:36]

LGTM, thanks

https://codereview.appspot.com/9810044/diff/1/tests/test_utils.py
File tests/test_utils.py (right):

https://codereview.appspot.com/9810044/diff/1/tests/test_utils.py#newcode98
tests/test_utils.py:98: agent file containing the addresses.
So I looked at the juju-core code and they strongly have the concept of a
'machiner', an agent that oversees the lifecycle of a machine.  So as odd as it
looks in this context I guess we should go with it.

[frankban, 2013-06-03 14:46:33]

*** Submitted:

Take the juju API address from the environment

The charm now tries to retrieve the juju API
address from the hook context before parsing
the machiner agent file.

This way we take advantage of a change in juju-core 
currently in review, which introduces the corresponding
functionality.

R=
CC=
https://codereview.appspot.com/9810044

https://codereview.appspot.com/9810044/diff/1/hooks/utils.py
File hooks/utils.py (right):

https://codereview.appspot.com/9810044/diff/1/hooks/utils.py#newcode135
hooks/utils.py:135: base_dir = os.path.abspath(base_dir)
On 2013/05/31 15:22:04, teknico wrote:
> Sorry, can't resist: :-)
> 
>     segs = (CURRENT_DIR, '..', '..') if unit_dir is None else (unit_dir, '..')
>     base_dir = os.path.abspath(os.path.join(*segs))

I agree this avoids some code duplication and it's cool, but I still consider my
version slightly more readable.

https://codereview.appspot.com/9810044/diff/1/tests/test_utils.py
File tests/test_utils.py (right):

https://codereview.appspot.com/9810044/diff/1/tests/test_utils.py#newcode98
tests/test_utils.py:98: agent file containing the addresses.
On 2013/06/03 13:08:36, bac wrote:
> So I looked at the juju-core code and they strongly have the concept of a
> 'machiner', an agent that oversees the lifecycle of a machine.  So as odd as
it
> looks in this context I guess we should go with it.

Ok.

[frankban, 2013-06-03 14:48:26]

Thanks fro the reviews Nicola and Brad!