Re: the standard function enumFrom

On Monday 25 October 2010 17:12:45, Paqui Lucio wrote:
> Hi,
> Given the following enumerated type (as an example)
>
> data P =3D A | B | C | D | E
>          deriving (Show,Enum)
>
> and the Prelude definition:
>  	    enumFrom x  =3D map toEnum [ fromEnum x ..]

That's not a definition to be used for all cases, it's a default definition=
=20
to be used if nothing better is provided.

>
> I guess that  the two expressions
> 	1) (map toEnum [ fromEnum A ..]) :: [P]
> 	2)  enumFrom A
> should produces the same result, say
> 	[A,B,C,D,E] :: [P]
>
> However, the expression 1)  produces an error as a consequence of trying
> to apply toEnum to the number 5.

Because [fromEnum A .. ] is [0 .. ], which is [0 .. maxBound :: Int] and=20
the derived Enum instance calls error for arguments of toEnum outside the=20
range [0 .. number of constructors - 1].

> =C2=BFWhy the expression 2) doesn't not produce the same error?

The derived Enum instance is cleverer, it uses the number of constructors=20
in the datatype to determine upper bounds for enumFrom and enumFromTo, so=20
that [x .. ] and [x, y .. ] don't try to call toEnum on an invalid=20
argument.

> =C2=BFDoes Hugs not apply the definition in the Prelude for evaluating the
> expression 2)?

No, the derived Enum instances don't use the default method (which would=20
rarely be the right thing to do).

> There should be something that I'm missing in this computation.
> Some hints?
>
> Thanks in advance,
> Paqui