Bug 744719

Created attachment 527200 [details]
using nested enums and enums on nested facts

Description of problem:
Dropdown list with enumeration values is missing, if you use a fact model with nested fact and enumeration for the nested fact's field.
I use the two "types" of enumerations - one defined on a fact that's nested into another fact in the model and one nested, defined directly on the topmost fact and listing values for the subfact's field - see below.


Model used:

  declare Subfact1
 	enumfield1: String
  end

  declare Subfact2
	enumfield2: String
  end

  declare Fact1
	subfact1field: Subfact1
  end

  declare Fact2
	subfact2field: Subfact2
  end

Enumeration used:

  'Fact1.subfact1field.enumfield1' : ["val1a", "val1b"]
  'Subfact2.enumfield2' : ["val2a", "val2b"]


When I try to use these facts in a guided rule, the dropdown enumeration list is not shown, i get a simple edit box instead. This applies to these cases:
1) Adding 'Fact1' and it's field 'subfact1field', then the field within that, 'enumfield1'. Selecting a literal value for 'enumfield1'.
2) Adding Fact1 and choosing Expression editor - then adding 'subfact1field' and
'enumfield1'. Selecting a literal value for 'enumfield1', again.
3) and 4) as above, but for 'Fact2' and corresponding fields.
5) Adding 'Fact1' and binding it to variable, let's say 'f1'. Then adding 'From' and choosing 'f1'. Adding a pattern of 'Subfact1' and its field 'enumfield1' and, again, matching it with a literal value.

Though 5) is a workaround for the issue if used with 'Fact2' and its fields.

See the screenshot if it doesn't seem clear.



Version-Release number of selected component (if applicable):
BRMS 5.2.0 ER5



Additional info:
by manstis:
> ...an unreported problem remains: using enumerations on nested models (e.g.
> ParentType.ChildType.enumField == XYZ). ... The problem is caused by
> enumerations being driven by the parent Fact type (which in part is to support
> linked enums, where the value of one depends on another). It is not a simple
> fix. ...there is a work-around and it is *possibly* an edge case.