Open Stack

On Tue, Apr 16, 2013 at 04:13:06PM +0200, Thomas Spatzier wrote:
> Hi Adrian,
> 
> thanks for taking writing this up to capture yesterday's discussions.
> I don't fully understand what the difference or overlap of the Model
> Interpreter inside the main blue box and the Model Interpreter in the green
> box is. My understanding was that a Model Interpreter would be something
> that translates an external format (such as CFN, TOSCA, whatever) to the
> core Heat DSL. And then there would be one component which looks at the
> common DSL model to derive the deployment plan (I think somebody said this
> is what the "parser" does today).
> I think having to partly re-implement the interpreter (in the sense of
> compiling a concrete deployment plan directly) for various external formats
> could be complicated.

I agree, which is why I propose keeping the mult-format translation outside
of heat (layered above the API, or some scripts which do format conversion)

The "parser" is really our internal terminology, but for the purposes of
the discussion, the parser is the thing which converts the native template
format into resource definitions and a dependency-graph we can process

Steve