In type definitions, type constructors are silently mistaken as data constructors

[ulidtko]

It's easy to make an error like this:

:- import_module list, int.

:- type myintlist ---> list(int).
:- func MyMerge(myintlist, myintlist) = myintlist.
MyMerge(A, B) = A ++ B.

The compiler says:

hello.m:013: In clause for function `'/3:
hello.m:013:   in function result term of clause head:
hello.m:013:   type error in unification of variable `HeadVar__4'
hello.m:013:   and functor `++/2'.
hello.m:013:   variable `HeadVar__4' has type `hello.myintlist',
hello.m:013:   functor `++/2' has type
hello.m:013:   `(list.list(T) ++ list.list(T)): list.list(T)'.

which is horrific! (and even worse for set). It gives no insight whatsoever.

Corrected code is:

:- type myintlist ---> mycon(list(int)).
:- func MyMerge(myintlist, myintlist) = myintlist.
MyMerge(mycon(A), mycon(B)) = mycon(A ++ B).

and, perhaps, experienced Mercury coder won't do this mistake — but newbies likely will.

My point is that it'd be easy to check if a user-defined data constructor name clashes with a type/type constructor name in scope, and issue a warning or suggestion for such a type definition. It'd help avoid the error.