environs/*: access environ name safely

environs/*: access environ name safely

As pointed out by Jeroen Vermeulen, the environment name
was set by SetConfig but accessed without
locking. This patch sets the field only once.
This is valid because the environment name must never change.

https://code.launchpad.net/~rogpeppe/juju-core/328-environs-safe-name/+merge/171058

(do not edit description out of merge proposal)

[rog, 2013-06-24 11:00:44]

Please take a look.

[jtv.canonical, 2013-06-24 15:36:46]

LGTM — but don't forget the openstack provider, which does the same thing!

Thanks for the fix.  I think it's simple and effective, and to be honest I'm
just a little bit surprised that it doesn't break any tests.  :)

I suspect the name is conventionally cached in the environment so as to allow
non-blocking access from call trees that may or may not have the lock at the
time.  It's one of those things that would be nice to have in a generalized
layer on top of the providers.

I can propose a similar fix for the Azure provider.

[thumper, 2013-06-24 22:29:10]

On 2013/06/24 15:36:46, jtv.canonical wrote:
> LGTM — but don't forget the openstack provider, which does the same thing!
> 
> Thanks for the fix.  I think it's simple and effective, and to be honest I'm
> just a little bit surprised that it doesn't break any tests.  :)
> 
> I suspect the name is conventionally cached in the environment so as to allow
> non-blocking access from call trees that may or may not have the lock at the
> time.  It's one of those things that would be nice to have in a generalized
> layer on top of the providers.
> 
> I can propose a similar fix for the Azure provider.

LGTM

[dave_cheney.net, 2013-06-25 04:04:33]

On 2013/06/24 22:29:10, thumper wrote:
> On 2013/06/24 15:36:46, jtv.canonical wrote:
> > LGTM — but don't forget the openstack provider, which does the same thing!
> > 
> > Thanks for the fix.  I think it's simple and effective, and to be honest I'm
> > just a little bit surprised that it doesn't break any tests.  :)
> > 
> > I suspect the name is conventionally cached in the environment so as to
allow
> > non-blocking access from call trees that may or may not have the lock at the
> > time.  It's one of those things that would be nice to have in a generalized
> > layer on top of the providers.
> > 
> > I can propose a similar fix for the Azure provider.
> 
> LGTM

LGTM.

[rog, 2013-06-26 13:36:21]

Please take a look.