On 20/11/2010 08:42 AM, Tillmann Rendel wrote:
Hi Andrew,
Andrew Coppin wrote:
Now, what about type variables? What do they do? Well now, that seems to
be slightly interesting, since a type variable holds an entire type
(whereas normal program variables just hold a single value), and each
occurrance of the same variable is statically guaranteed to hold the
same thing at all times. It's sort of like how every instance of a
normal program variable holds the same value, except that you don't
explicitly say what that value is; the compiler infers it.
What do you mean by "hold the same thing at all times"?
Consider the following program:
id :: forall a . a -> a
id x = x
call1 :: Bool
call1 = id True
call2 :: Int
call2 = id 42
This program contains a type variable a, and a value variable x. Now,
these variables do *not* mean the same thing at all times. In the
first call of id, a is Bool and x is True; but in the second call of
id, a is Int and x is 42. If these variables would mean the same thing
at all times, I would expect them to be called constants, wouldn't you?
That would be why I said "instance of a variable", rather than
"variable". (There's also the alpha conversion problem, for example.)
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
