diff: typeclass bug fix

[David Glen JEFFERY]

Hi,

Could someone (Fergus?) review this?

-----------------------------------------------------------------------------

Estimated hours taken: 4

Fix a bug where polymorphism.m was erroneously applying a type substition
recursively, causing an infinite loop in certain situations.

compiler/polymorphism.m:
	Change a call to term__apply_rec_substitution_to_list into
	term__apply_substitution_to_list.

	Also put the goals to initialise the superclass type-infos *before*
	the construction unification for the sub-class.

-----------------------------------------------------------------------------

Index: polymorphism.m
===================================================================
RCS file: /home/staff/zs/imp/mercury/compiler/polymorphism.m,v
retrieving revision 1.136
diff -u -t -r1.136 polymorphism.m
--- polymorphism.m	1998/06/09 02:14:19	1.136
+++ polymorphism.m	1998/06/17 09:38:53
@@ -1149,7 +1149,7 @@
         ClassId = class_id(ClassName, ClassArity),
         term__vars_list(NewConstrainedTypes, NewConstrainedVars),
         list__append(NewConstrainedVars, ConstrainedVars0, ConstrainedVars),
-        term__apply_rec_substitution_to_list(NewConstrainedTypes, TypeSubst, 
+        term__apply_substitution_to_list(NewConstrainedTypes, TypeSubst, 
                 ConstrainedTypes0),
         % we need to maintain the invariant that types in class constraints
         % do not contain any information in their term__context fields
@@ -1534,7 +1534,7 @@
 
         TypeClassInfoGoal = Unify - GoalInfo,
         NewGoals0 = [TypeClassInfoGoal, BaseGoal],
-        list__append(SuperClassGoals, NewGoals0, NewGoals),
+        list__append(NewGoals0, SuperClassGoals, NewGoals),
         Info = poly_info(VarSet, VarTypes, TVarSet, TVarMap, 
                         TCVarMap, Proofs, PredName, ModuleInfo).
 
-----------------------------------------------------------------------------

New file: tests/hard_coded/typeclasses/instance_superclass.m:

:- module instance_superclass.

% This is a regression test for the case where there is a superclass
% relationship, and the instance declaration has a type variable in the
% instance type.

:- interface.

:- import_module list.

:- typeclass c1(T) where [ ].
:- typeclass c2(T) <= c1(T) where [ ].

:- instance c1(list(T)) where [ ].
:- instance c2(list(T)) where [ ].

:- pred p is det.

:- implementation.

	% The bug that this test case is checking for is at the creation of
	% the typeclass info for this call: if the substitution from "T" in
	% the typeclass decl to "list(T)" in the instance is applied
	% recursively, an infinite loop results.
p :- q([1]).

:- pred q(T) <= c2(T).
:- mode q(in) is det.

q(_).

-----------------------------------------------------------------------------

love and cuddles,
dgj
-- 
David Jeffery (dgj at cs.mu.oz.au) |  Marge: Did you just call everyone "chicken"?
MEngSc student,                 |  Homer: Noooo.  I swear on this Bible!
Department of Computer Science  |  Marge: That's not a Bible; that's a book of
University of Melbourne         |         carpet samples!
Australia                       |  Homer: Ooooh... Fuzzy.