[Haskell] Why is newChan in the IO Monad?

[Alastair Reid]

On Friday 23 April 2004 20:05, S. Alexander Jacobson wrote:
> Yes, that makes sense, but I'm ok with passing in
> an identity.  I'd like a function like this:
>
>   newChanSafe::Identity -> Chan a
>   type Identity = Double -- or whatever

As Nick observes, using this function would require you to pass around a 
supply of unique Identitys.  If we assume you're going to have to do this, 
why not simplify things and pass around a list of newChans:

  type Identity = Chan Int
  withChan :: (Identity -> a) -> [Identity] -> (a,[Identity])
  ...

You can use unsafeInterleaveIO to create a lazy list of channels.
Better yet, you can use unsafeInterleaveIO to create a lazy
 tree of channels so that splitting the supply is efficient.


One minor detail left as an exercise: my definition of Identity is monomorphic 
but you probably want them to be polymorphic.  This is a bit tricky to fix 
and will require a monad (or equivalent) to ensure that you don't allocate 
the same chan twice and then use it at different types.  (Probably requires 
unsafeCast too.)

--
Alastair Reid