Open Stack

On 08/17/2014 05:11 AM, Stan Lagun wrote:
>
> On Fri, Aug 15, 2014 at 7:17 PM, Sandy Walsh <sandy.walsh at rackspace.com
> <mailto:sandy.walsh at rackspace.com>> wrote:
>
>     I recently suggested that the Ceilometer API (and integration tests)
>     be separated from the implementation (two repos) so others might
>     plug in a different implementation while maintaining compatibility,
>     but that wasn't well received.
>
>     Personally, I'd like to see that model extended for all OpenStack
>     projects. Keep compatible at the API level and welcome competing
>     implementations.
>
>
> Brilliant idea I'd vote for

The problem is when the API is the worst part of the project.

We have a number of projects (some that I work on) that one of the 
weakest parts of the project is the design, inconsistency, and 
efficiency of the API constructs are simply terrible.

The last thing I would want to do is say "here, everyone go build 
multiple implementations on top of this crappy API". :(

As for the idea of letting the market flush out competing 
implementations, I'm all for that ... with some caveats. A couple of 
those caveats would include:

  a) Must be Python if it is to be considered as a part of OpenStack's 
integrated release [1]
  b) The API must be simple, efficient, and consistent, possibly having 
signoff by some working group focused on API standards

All the best,
-jay

[1] This isn't saying other programming languages aren't perfectly 
fine*, just that our integration and CI systems are focused on Python, 
and non-Python projects are a non-starter at this point.

* except Java, of course. That goes without saying.