[mercury-users] float__min

[Tyson Dowd]

On 07-Jan-1999, Mark NG <markn at students.cs.mu.oz.au> wrote:
> 
> G'day !
> 
> I am rather new to mercury and having troubling using float__min (func/0).
> 
> I tried to compile this program :-
> 
> :- module float_test.
> :- import_module io.
> :- interface.
> :- pred main(io__state::di, io__state::uo) is det.
> %--------------------------------------------------------------------%
> :- implementation.
> :- import_module float.
> main -->
>         { F = float__min },
>         io__print(F).
> 
> %------------------------------------------------------------
> and got...
> 
> float_test.m:015: In clause for predicate `float_test:main/2':
> float_test.m:015:   error: ambiguous overloading causes type ambiguity.
> float_test.m:015:   Possible type assignments include:
> float_test.m:015: F :: float or (func(float, float) = float)
> 
> 
> How do I get it to work ??

There are two functions called `float__min' -- you need to make
sure the compiler can figure out which one you mean.

The only thing you do with F is print it.
io__print is pretty generic -- it doesn't provide any extra type information.

If you add a goal such as
	{ _X = F + 0.0 }
this will clearly mark F as being of type `float' (unless you define
a `+' that works on things of type `func(float, float) = float').
But it's a bit ugly.

You could use a print statement such as io__write_float instead of
io__print -- this will make it clear that F is of type `float'

Or you can write a predicate is_float(float::in) which is always
true, and use it on F to make sure it is a float.

--
Tyson.