[mercury-users] Destructive map_replace

[Julien Fischer]

On Mon, 5 Jan 2009, Michael Day wrote:

> Here is a C implementation of map_replace, changed to return a boolean result 
> instead of a maybe, to simplify the C code and reduce allocation:
>
> :- pred replace_foo(string, list(string), bool).
> :- mode replace_foo(ui, uo, out) is det.
>
> replace_foo(X, Xs, Res) :-
>    ( if X = "foo" then
>        copy(["bar"], Xs),
>        Res = yes
>    else
>        Xs = [],
>        Res = no
>    ).
>
> :- pred call_pred(pred(T, list(T), bool), T, list(T), bool).
> :- mode call_pred(in(pred(ui, uo, out) is det), ui, uo, out) is det.
>
> call_pred(Pred, X, Xs, Res) :- Pred(X, Xs, Res).
>
> :- pragma foreign_export(c, call_pred(in(pred(ui, uo, out) is det), ui, uo, 
> out), "call_pred").
>
> :- pred map_replace(pred(T, list(T), bool), list(T), list(T)).
> :- mode map_replace(in(pred(ui, uo, out) is det), di, uo) is det.
>
> :- pragma foreign_proc(c,
>        map_replace(Pred::in(pred(ui, uo, out) is det), Xs::di, Ys::uo),
>        [will_not_call_mercury, thread_safe, promise_pure], "
>    {
>        MR_Word *AddrOfPrev;
>
>        Ys = Xs;
>        AddrOfPrev = &Ys;
>
>        while (!MR_list_is_empty(Xs))
>        {
>            MR_Integer X = (MR_Integer) MR_list_head(Xs);
>            MR_Word Xs0 = (MR_Word) MR_list_tail(Xs);
>            MR_Word NewXs;
>            MR_Word Res;
>            MR_Word Fake;
>
>            call_pred(Fake, Pred, X, &NewXs, &Res);
>
>            if (Res != 0)
>            {
>                if (MR_list_is_empty(NewXs))
>                {
>                    *AddrOfPrev = Xs0;
>                }
>                else
>                {
>                    *AddrOfPrev = NewXs;
>
>                    do
>                    {
>                        Xs = NewXs;
>                        NewXs = (MR_Word) MR_list_tail(NewXs);
>                    }
>                    while (!MR_list_is_empty(NewXs));
>
>                    MR_list_tail(Xs) = Xs0;
>
>                    AddrOfPrev = &(MR_list_tail(Xs));
>                }
>            }
>            else
>            {
>                AddrOfPrev = &(MR_list_tail(Xs));
>            }
>
>            Xs = Xs0;
>        }
>    }
> ").
>
> One question: I'm passing a fake type info to call_pred, and nothing seems to 
> go wrong yet, but I feel guilty. Should I be?

When the first argument of call_pred needs to use RTTI you will be ;-)

Julien.
--------------------------------------------------------------------------
mercury-users mailing list
Post messages to:       mercury-users at csse.unimelb.edu.au
Administrative Queries: owner-mercury-users at csse.unimelb.edu.au
Subscriptions:          mercury-users-request at csse.unimelb.edu.au
--------------------------------------------------------------------------