union_data of unmatching domains (but equal size) should still give an error, right?

[jipclaassens]

Here, union_unit of Noord-Holland and Zuid-Holland, but union_data uses Noord-Holland twice. This gives no error. Both Noord and Zuid-Holland have five rows.

...\GeoDMS-Test\Operator\cfg\Operator.dms
/Relational/UnTiled2Tiled/union_data/CityNames_shouldbeerror

unit<uint32>    unit_ProvCity        := union_unit(Source/NoordHollandBigCities, Source/ZuidHollandBigCities);

attribute<string> CityNames_shouldbeerror    (union_unit/unit_ProvCity) := union_data(union_unit/unit_ProvCity, Source/NoordHollandBigCities/CityNames, Source/NoordHollandBigCities/CityNames);

[MaartenHilferink]

Not a defect: union_data is designed to be used to relabel domains.
To check and/or unify the domain of the result of a union_data with another domain, for example:

• the domain of another instance of union data
• the result of a union_unit
• other?

one has to impose criteria that might be too restrictive in, for example:

1. X := combine_unit(A, B) and ='union_data(X, '+ asItemList(A/name+'/attr_per_B') +')';
2. ='union_data(B, '+ asItemList(B/name+'/specific_parameter') +')';
3. (more difficult to validate examples here?)

Note however, that 1. could be modelled as combine_unit(A, B), sort of is equivalent with ='union_unit('+Repeat('B', NrOfRows(A))+')'
and 2.: B is equivalent with ='union_unit( '+ asItemList('domainUnit(B/V/'+B/Name+')')+')';

which could have been considered if meta-scripting somehow annotated the resulting primary calculation rules.