[sc34wg3] More on TMCL MAX_INT

[Lars Marius Garshol]

Got this reply from Kal offlist:

> IMO * is more intuitive than INF for cardinality constraints.  
> Especially for those familiar with modelling in UML or in Protege.  
> Even "unbounded" as seen in XML Schema would be preferable to INF.

There's no question that "*" is more intuitive, but technically, what  
it would have to mean is "INF"^^xsd:double, because it's a literal  
value going into the card-max occurrence. This would be part of the  
CTM syntax, and not something TMCL-specific.

"unbounded" could of course be treated the same way.

> Also I think that using double as your data type is a mistake  
> because any non-integer values are nonsense/invalid.

Well, this would be a kind of compromise where all values must be  
either xsd:integer *or* "INF"^^xsd:double.

The reason INF is xsd:double and not xsd:integer is that xsd:integer  
does not have an INF value. INF is only a value of xsd:double, and not  
even of xsd:decimal.

> You missed out of your list of effects that the processor now has to  
> do some special case validation of the datatype, which in effect  
> balances out the work caused by defining a new, separate data type.

Yes. That's why I don't like this proposal. Note that "*" and  
"unbounded" amount to the same thing, unless we invent our own  
datatype. Whatever we do, none of these values are valid xsd:integers.

Anyway, I prefer cardinality topics.

--Lars M.
http://www.garshol.priv.no/blog/
http://www.garshol.priv.no/tmphoto/