Should Property facet's DataType be made optional?

[andyward]

This was triggered by #420 and ultimately revisits #206 and a number of other issues around property dataTypes.

The IDS1.0 decision to make dataType a mandatory field on properties when specifying a value has various UX and functional knock-on impacts including

1. IDS Authors and end users often don't know/care what datatype a value uses - they just want to specify a value/range (often just a text literal)
2. Being too prescriptive produces false negatives when software outputs data types that are slightly different (i.e. IfcText vs IfcLabel, IfcLengthMeasure vs IfcPositiveLengthMeasure) Data type mismatch in applicability bypasses checking #337 / Proposal: Permit dataType equivalents for non-measure Property values #331
3. Because dataType is a string literal and can't support complex constraints, IDS Authors may need to duplicate specs when checking values (e.g. to include IfcText and IfcLabel variants)
4. Prohibiting property values in the requirement is not supported. So you can't say "InstallationDate should not have a value of 'n/a'" Invalid cardinality 'prohibited' on property, it is not compatible with the specification of the 'dataType' constraint. #420 (without expressing as different construct using prohibited applicability)

The last one in particular seems to be unintuitive for users. I can prohibit a property by Name / Pset, but not by it's value?

In IDS future I'd propose DataType becomes optional. It would also make sense if it could support complex constraints so we could have enums, regex wildcards etc.

Not sure what the counter argument may be - But I suppose there are examples where we need to be really clear on the dataType/measure being used, such as 'Nominal Length' properties must be IfcLengthMeasures (not IfcText). By making DataType optional we can then be prescriptive when it is important.