classes

[Juan Carlos Arévalo Baeza]

On Fri, 31 Aug 2001 12:56:32 +0300, Cagdas Ozgenc wrote:

>I am a newbie, so I might be making =A0nonsense.

   You actually make much sense, and you betray your OO=
 background (C++, maybe?). Programming in Haskell requires a very=
 different way of thinking.

>In the following definiton:
>
>class Eq a where =A0 =A0 (=3D=3D) :: a ->=A0a ->=A0 Bool
>
>Why do we conlude that both of the "a" refer to the =A0same type?

   In Haskell, a class is a very different concept than what you=
 find in OO languages like C++. In OO languages, a class is the=
 definition of a data type and its operations. In Haskell, it is=
 an abstract set of operations but without a concrete data type.=
 So:

class Eq a where =A0 =A0 (=3D=3D) :: a ->=A0a ->=A0 Bool

   indicates that if a data type 'a' has been made an instance of=
 'Eq', then that data type 'a' has the operation (=3D=3D) as defined=
 in that instance. For example, if I define a data type:

data MyEnum =3D Sel1 | Sel2 | Sel3

   and then define the instance of Eq:

instance Eq MyEnum where
   Sel1 =3D=3D Sel1 =3D True
   Sel2 =3D=3D Sel2 =3D True
   Sel3 =3D=3D Sel3 =3D True
   _    =3D=3D _    =3D False

   then it's the same as having defined, in C++, a function such=
 as:

bool operator=3D=3D(MyEnum s1, MyEnum s2)
{
   if (s1 =3D=3D Sel1 && s2 =3D=3D Sel1) return true;
   if (s1 =3D=3D Sel2 && s2 =3D=3D Sel2) return true;
   if (s1 =3D=3D Sel3 && s2 =3D=3D Sel3) return true;
   return false;
}

   One advantage is that, in Haskell, you can specifically assert=
 for instances of classes, while in C++ you have to rely in=
 compiler errors that can be disorienting at times. For example,=
 std::set<val> requires that the parameter type have a '<'=
 operator, but you can't assert on it. In Haskell you can, using=
 contexts:

class Ord a where
   (<) :: a -> a -> Bool

data Set a =3D Ord a =3D> [...]

   I don't know if I'm making sense. I'm not that proficient in=
 Haskell myself.

>Apparently using different types that both belonging to class=
 Eq
>results in an error. For example
>
>65 =3D=3D 'a'

   As Ketil said, this can be done only using a non-standard=
 extension of GHC. I believe there are a lot of people that wish=
 it were standard, so I guess it'll be some day.

>It doesn't make much sense at first, however, how =A0do you go by
>implementing multiple polymorphism?

   There are several kinds of polymorphism. For example, you=
 could say Haskell supports multiple type polymorphism (as in=
 multiple template arguments in C++).

>Secondly, why does Haskell distinguish between a =A0Type and a=
 Class?
>There seems to be no benefit of such approach.

   Oh, there's benefit. It does require a serious change in=
 perspective, though. It can't be so easily explained.

>Thirdly, is it possible to define meta-classes in =A0Haskell?

   Could you give an example of what you mean? If you mean=
 meta-programming, no, Haskell doesn't really have any of that.=
 You can do some things using polymorphic types, and the fact=
 that Haskell is pure functional means that often the line=
 between compile-time and run-time computations can blur a lot.
 
   Salutaciones,
                        JCAB
email: jcab@roningames.com
ICQ: 101728263
The Rumblings are back: http://www.JCABs-Rumblings.com