Naming of Config keys

[Julian Edwards]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 26/06/13 18:51, John Arbash Meinel wrote:
> This just came up here: 
> https://code.launchpad.net/~rvb/juju-core/az-public-storage/+merge/171251
>
>  It came up when we did the Openstack provider, but we didn't have
> a public discussion of it.
> 
> Is it better to have config keys named the same between
> environments, or is it better to have the config keys make the most
> sense for a given environment.
> 
> For example, ec2 calls S3 things Buckets, but Openstack calls them 
> Containers, as does Azure.
> 
> We have a config for "public-bucket" which defines the name of the 
> bucket-like-thing we want to look in for tools.
> 
> When we wrote the Openstack code, we originally called it 
> "public-container", but were told we should go for consistency
> with ec2. It came up again for Azure (because people implementing
> the code for X naturally feel things should be named after the
> thing their immediately working on).
> 
> I think the goal with cross-environment naming of keys is that we 
> wanted it to be as easy as possible to move your workload from one 
> provider to another.
> 
> I'm a bit conflicted, because I personally prefer the
> names-in-context that Raphael is proposing, but we went with the 
> cross-environment-naming on request. I just want to make sure that 
> request still stands.
> 
> John =:->
> 

I like consistency, but "bucket" means far less to me than
"container".  Therefore I don't see the problem in having
provider-specific config names for similar concepts, since there's
going to be  other config items totally unique to each one anyway.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iEYEARECAAYFAlHLoAMACgkQWhGlTF8G/HchBACZAWtaxy1ScXWjB9xUC59t4P/l
fBkAn0/leFP69EldzTD/MW1oYTV6QDdP
=GLbs
-----END PGP SIGNATURE-----