[m-rev.] Re: for review: parsing_utils improvements

[Ian MacLarty]

On Tue, Sep 29, 2009 at 12:50 PM, Ian MacLarty
<maclarty at csse.unimelb.edu.au> wrote:
> Index: library/parsing_utils.m
> ===================================================================
...
> +
> +parse(InputString, SkipWS, Parser, Result) :-
> +    % This is pure, because it will always return the same results for
> +    % the same inputs (the mutable in Src cannot be accessed outside
> +    % of the promise_pure scope below).
> +    promise_pure(
> +        new_src_and_ps(InputString, SkipWS, Src, PS0),
> +        ( Parser(Src, Val, PS0, _) ->
> +            Result = ok(Val)
> +        ;
> +            MutVar = Src ^ error_info,
> +            semipure get_mutvar(MutVar, error_info(Offset, MaybeMsg)),
> +            offset_to_line_number_and_position(src_to_line_numbers(Src),
> +                Offset, Line, Col),
> +            Result0 = error(MaybeMsg, Line, Col),
> +            % We make parse/4 cc_multi because declaratively
> +            % parse(Str, SkipWS, Parser, error(MaybeMsg, Line, Col)) is true
> +            % for all MaybeMsg, Line and Col iff
> +            %   new_src_and_ps(Str, SkipWS, Src, PS0),
> +            %   Parser(Src, _, PS0, _)
> +            % is false, but operationally MaybeMsg, Line and Col are
> +            % restricted to one value each.
> +            cc_multi_equal(Result0, Result)

Peter Wang and have been discussing whether this should be cc_multi
and we both think it should actually be det.  The reason is that the
determinism is an operational property of the procedure and not a
declarative property.  From the reference manual:

For each mode of a predicate or function,
we categorise that mode according to how many times it can succeed,
and whether or not it can fail before producing its first solution.

So, I propose to make the parse/4 predicate det.  Are the any objections?

By the same reasoning I think that exception.try should also be det
and not cc_multi.  Would there be any objections to changing that? (in
a separate diff of course)

Ian.

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