Re: The rationale of well-formed-ness (Re: [IMPORTANT] Necessary revision of the sugar proposal)

Hi PA,
remember that I am in the context of "my" revised sugar proposal, which has the following property:
(1) RDF 1.1 graphs are always well-formed, and the RDF-star extension does not play any role whatsoever;
(2) if RDF-star graphs systematically use the syntactic sugar expression for reification, then entailments are always well-formed - in other words, by avoiding the use of rdf-star:is-reification-of one is guaranteed to be always "well-formed".

Apparently, you consider that such a notion of "soft" constraint is a bad idea. Fair enough. But then could we please call hard constraints a different name? In fact, do we need a name for them? They just become part of the abstract syntax...

I believe that if rdf-star introduces a new construct (i.e., rdf:is-reification-of) it is pretty natural to state the only cases where it could be used. Any other use of this predicate in a triple would be deprecated; it is only about the newly introduced predicate.

In fact, if we need to add something to the abstract syntax, I think the two options that were discussed last week ("triple terms" of "edge statements") are much more elegant than constraining the use of a bunch of predicates... But maybe that's what you are getting at? :->

I would like you to get to that, but I am in no position to push in this direction. But yes, this would be an elegant solution. But note that this would not solve the issue of needing well-formedness, if you want to remain in a "sugar" approach. Of course, by leaving the sugar approach and just having edge declarations, then we do not need well-formedness at all.

NB: I have ideas for (possibly) better names for "well-formed-ness", if the term is deemed misleading. But I don't want to derail the conversation with bikeshedding.

Maybe "meaning-preserving"?
--e.