[m-dev.] for review: simplify.m bug fix

[Fergus Henderson]

On 30-Aug-1998, Simon Taylor <stayl at cs.mu.OZ.AU> wrote:
> Clean up the handling of commits and goals which cannot succeed.
> This fixes problems with tests/hard_coded/cut_test.m.
...
> Index: compiler/deforest.m
> ===================================================================
> RCS file: /home/staff/zs/imp/mercury/compiler/deforest.m,v
> retrieving revision 1.4
> diff -u -t -u -r1.4 deforest.m
> --- deforest.m	1998/07/08 20:55:53	1.4
> +++ deforest.m	1998/08/11 06:31:39
> @@ -1585,14 +1585,16 @@
>                  pd_info_get_size_delta(SizeDelta0),
>                  pd_info_get_changed(Changed0),
>  
> -                { Goal0 = _ - GoalInfo0 },
> -                { Goal1 = GoalExpr1 - GoalInfo1 },
> -
> -                        % Take the non-locals from the calling goal_info,
> -                        % everything else from the called goal_info.
> -                { goal_info_get_nonlocals(GoalInfo0, NonLocals0) },
> -                { goal_info_set_nonlocals(GoalInfo1, NonLocals0, GoalInfo2) },
> -                { Goal2 = GoalExpr1 - GoalInfo2 },
> +                        % update the quantification if not all the output
> +                        % arguments are used.
> +                { Goal1 = _ - GoalInfo1 },
> +                { goal_info_get_nonlocals(GoalInfo1, NonLocals1) },
> +                { set__list_to_set(Args, NonLocals) },
> +                ( { NonLocals1 = NonLocals } ->
> +                        pd_util__requantify_goal(Goal1, NonLocals, Goal2)
> +                ;
> +                        { Goal2 = Goal1 }
> +                ),

Shouldn't that be `NonLocals1 \= NonLocals'?

Also, you should use `set__equal' rather than `=' to compare sets
for equality (although as it happens with the current representation
of sets this doesn't matter).

> +simplify__goal_2(conj(Goals0), GoalInfo0, Goal, GoalInfo, Info0, Info) :-
>          simplify_info_get_instmap(Info0, InstMap0),
>          simplify__conj(Goals0, [], Goals, GoalInfo0, Info0, Info1),
> -        simplify_info_set_instmap(Info1, InstMap0, Info),
> -        ( Goals = [SingleGoal] ->
> +        simplify_info_set_instmap(Info1, InstMap0, Info2),
> +        ( Goals = [] ->
> +                goal_info_get_context(GoalInfo0, Context),
> +                true_goal(Context, Goal - GoalInfo),
> +                Info = Info2
> +        ; Goals = [SingleGoal - SingleGoalInfo] ->
>                  % a singleton conjunction is equivalent to the goal itself
> -                SingleGoal = Goal - _
> +                simplify__maybe_wrap_goal(GoalInfo0, SingleGoalInfo, 
> +                        SingleGoal, Goal, GoalInfo, Info2, Info)
...
> +simplify__goal_2(par_conj(Goals0, SM), GoalInfo0, Goal,
> +                GoalInfo, Info0, Info) :-
>          (
>                  Goals0 = []
>          ->
> -                Goal = conj([]),
> +                goal_info_get_context(GoalInfo0, Context),
> +                true_goal(Context, Goal - GoalInfo),
>                  Info = Info0
>          ;
>                  Goals0 = [Goal0]
>          ->
> +                GoalInfo = GoalInfo0,
>                  simplify__goal(Goal0, Goal - _, Info0, Info)
>          ;

Why aren't singleton parallel conjunctions handled in the same way
as singleton ordinary conjunctions?

Otherwise that change looks OK.

-- 
Fergus Henderson <fjh at cs.mu.oz.au>  |  "I have always known that the pursuit
WWW: <http://www.cs.mu.oz.au/~fjh>  |  of excellence is a lethal habit"
PGP: finger fjh at 128.250.37.3        |     -- the last words of T. S. Garp.