Re: 5.1.6 Rule language coverage <--: UCR Requirements Text

Okay, how about this, for the coverage requirement?  I went for
verbosity/clarity in this version.  It can probably be cut to about the
last sentence, if the terminology is set up elsewhere.

     Because of the great diversity of rule languages, no one
     interchange language is likely to be able to bridge betweem all.
     Instead, RIF provides dialects which are each targetted at a
     cluster of similar rule languages.  Within that cluster, each
     feature of each rule language will have some degree of commonality
     with corresponding features of other rule languages in that
     cluster.  The RIF dialect targetting a cluster must support, at a
     minimum, interchange of rules using all the features which are
     common to all the major rule languages in that cluster. 

Does that do it?        (Say "yes", please!   :-)

      -- Sandro

> Sandro - good qu, but unfortunately (!) I don't have any underlying
> agenda other than (a) to justify RIF we need to make it successful and
> (b) that means covering the majority of rule engines in use with rules
> that would benefit from interchange.=20
> 
> Unfortunately also we cannot control (b) as it requires all major
> vendors to step up to the plate and do translators.=20
> 
> I think the main issue here is that you (justifiably) want the
> requirement to be measurable. So "coverage" really means support for
> rules in rule engines that are candidates for interchange. Can we get
> more detailed than that? Well of course that's where we need end-user
> use cases (like the MISMO example, and probably organisations like XBRL
> and FixML etc). But I'd be happy with taking MISMO-type rules (i.e.
> decision tables for data defined via an XML schema) as a starting point
> requirement.
> 
> In other words:=20
> - The "Coverage" critical success factor could be met by the adoption by
> 1 or more other technology / domain specific standards
> - An example requirement that would (provide some) support (for)
> standards like MISMO and PMML would be a decision table.
> 
> 
> Sorry to be obtuse!=20
> 
> Paul Vincent
> TIBCO | Business Optimization | Business Rules & CEP
> =20
> 
> > -----Original Message-----
> > From: Sandro Hawke [mailto:sandro@w3.org]
> > Sent: 06 June 2008 16:09
> > To: Paul Vincent
> > Cc: Chris Welty; Christian de Sainte Marie; RIF WG
> > Subject: Re: 5.1.6 Rule language coverage <--: UCR Requirements Text
> >=20
> >=20
> > > Sounds OK to me.
> > >
> > > [The thought occurs to me that "coverage" could be considered a
> > > critical-success-factor - which also translates directly into
> coverage
> > > as a requirement ie for RIF to concentrate on rule systems that are
> > > adopted and in use. Ignoring any Heisenberg uncertainty principle
> > > equivalent etc...]
> >=20
> > Yeah, it seems to me that this kind of thing (any "should" statement)
> is
> > a goal/CSF, not a requirement.  If you can't tell whether you've met a
> > requirement, what good is it?  Also, I think we should only accept
> > requirements we reasonably expect we can meet.
> >=20
> > I guess there's something powering this discussion, but I don't know
> > what.   Paul, what is it you want RIF-WG to do, in the days to come,
> > that you're getting at with this requirement?
> >=20
> >      -- Sandro

Received on Friday, 6 June 2008 16:05:42 UTC