Incorrect error about feature directions in inverse feature group

[rinsley]

I'm getting an error in this snippet of AADL in the AADL text editor:

feature group OutputGroup
	features
		Out1: out feature;
end OutputGroup;

feature group OutputGroupInverse inverse of OutputGroup
end OutputGroupInverse;

system Sys
	features
		IG1: in feature group inverse of OutputGroup;
		IG2: in feature group OutputGroupInverse;
end Sys;

IG1 and IG2 should both be valid as "in" direction groups, but IG2 is giving the error "All ports, parameters, feature groups, and abstract features in the referenced feature group type must satisfy the direction specified in the feature group."

[lwrage]

There are various edge cases where it is unclear what should be considered a valid model, e.g., feature groups without a type, feature groups with an empty type.
Also, should access features in a feature group be ignored in the validation, or should they result in an error?