Re: Extensibility: Fallback vs. Monolithic

Sandro Hawke wrote:
> Gary Hallmark <gary.hallmark@oracle.com> writes:
>   
>
>> I think things that can be ignored are metadata by definition, and we 
>> should get on with defining metadata.
>> I think the "dialect name" is metadata. 
>>     
>
> Is there any reason for a dialect name?  What would it be used for?
>   
its "just metadata", like rule and ruleset names, provenance, versions, 
comments, etc.  Should have no bearing on semantics. 
Maybe a RIF translator that supports only Core gets a ruleset document 
in a PR dialect.  Optimistically it scans it but unfortunately finds a 
"retract" syntax element.  It issues an error message like "can't 
process this *Production Rule* dialect because it has a retract in it".
>
> Well, the plan outlined in the charter is that in Phase 1 we demonstrate
> that Core can be extended to include any extension anyone cares about
> very strongly.  In theory, that's a lot easier than actually
> standardizing each of those extensions, but in practice maybe it's not.
>
>   
Hopefully, having an ASN that is easy to extend and that provides backward compatibility when materialized as an XML schema will be good enough.  I'm not optimistic about the possibility of discovering a way to have useful fallbacks nor extensions to the Core model theory for arbitrary extensions.  Especially not for Phase 1.