curious effect with generics

[forsyth]

Given

procedure Gen3 is
	generic
		X: Integer;
	package Foo_G is
		subtype T is Integer range 1 .. X;  --- <-(A)
	end Foo_G;

	package Foo_10 is new Foo_G(X => 10);

	package Foo_20 is new Foo_G(X => 20);
begin
	null;
end Gen3;

when traversing the generic declaration Foo_G -- not the Foo_10/Foo_20 instantiations but the original generic decl -- if (in the Ada interface) I call P_Is_Static_Expr on the node for the Identifier X in the range specification at (A), it returns True, and P_Eval_As_Int returns 20. If I comment out the declaration of Foo_20, the node X remains static and P_Eval_As_Int returns 10. If I comment out both Foo_x declarations, node X is not static. Is a binding created during the instantiations being left behind in the original generic declaration?

I should add that I first noticed it in connection with instantiations of a subprogram with a formal type parameter, where again it latched on to the type provided by the last appearing instantiation, but this example seemed even simpler.

[forsyth]

My impression is that recent changes to libadalang have addressed this notably change 11.50, so I'll close it for you for the time being.