Open Stack

On Tue, Apr 8, 2014 at 10:01 AM, Julien Danjou <julien at danjou.info> wrote:
> On Tue, Apr 08 2014, Doug Hellmann wrote:
>
>> I would like for us to continue to use the oslo prefix in some cases,
>> because it makes naming simple libraries easier but more importantly
>> because it is an indicator that we intend those libraries to be much
>> more useful to OpenStack projects than to anyone else. For projects
>> where that isn't the case (cliff, stevedore, taskflow, tooz, etc.) we
>> are already choosing "non-branded" names.
>
> I understand that, but can you point really a function that is
> so-damn-OpenStack-specific that if somebody would stumble upon it it
> would be like "pffff, what the hell!"? I don't think so. :)

A good bit of what we have is legacy code from choices made early in
nova's history. I wouldn't expect anyone else to use oslo.config
(ConfigObj is simpler) or oslo.rootwrap (sudo). The log configuration,
ContextAdapter, and ContextFormatter stuff we have in the log module
is fairly specific to setting up logging for our apps. Someone else
*could* use it, but I don't know why they would.

Some of it is newer, but still meant to share code between OpenStack
projects. The VMware team says oslo.vmware isn't useful outside of
OpenStack. The scheduler code in the incubator is another example of
this, as are some of the wrappers around things like the datetime and
uuid modules.

So it's fair to say we should look at each library as it graduates and
decide how to treat it, but I think the default for most of what we
have in the incubator right now is going to be an oslo.* library.

Doug

>
> --
> Julien Danjou
> // Free Software hacker
> // http://julien.danjou.info