Re: [TED] Action-188, ISSUE: production rule systems have "difficulty" with recursive rules in RIF Core

> > How do you suggest that I -- a user -- know what rules I can write and
> > remain confident they will work on several other vendors' systems?

> Sandro:
>   First of all, the poor user is the last person who will know : 
> ( However, the rule vendor should be able to communicate properly to  
> the user.
>   Secondly, the simple answer to the question is you cannot guarantee  
> that your rules will work on anybody else's engine. That is a  
> different project: not a rules interchange format but a standard  
> rules language. The trouble with the latter project is not that there  
> have not been any, but that there are too many! And that too many  
> people believe they know what such a SRL  should be like (including  
> yours truly)
>   Under the remit of the RIF, the best that you can hope for is that  
> you get a clear and unambiguous signal that your rule set has been  
> faithfully transmitted to a particular rule language. That is already  
> an advance on the state of the art, by the way.
> Frank

Why can't we transmit non-recursive Horn rules in a way that supports
cross-rule-platform portability?  I don't see the problem.

(on that note, I need to end this wonderful conversation until
tomorrow!) 

      -- Sandro

Received on Sunday, 17 December 2006 23:44:35 UTC