[Haskell] Re: A question about fundeps <-> GADT interaction

[Tomasz Zielonka]

On Fri, Dec 23, 2005 at 12:10:07PM +0100, Tomasz Zielonka wrote:
> On Fri, Dec 23, 2005 at 03:34:40AM -0000, oleg at pobox.com wrote:
> > 
> > [Sorry for possible duplication, our DNS server seems to be broken,
> >  and the sysadm is on vacation]
> > 
> > I don't think that is the problem with GADTs. The following works
> > 
> > > untype :: Term f a -> Term Untyped ()
> > > untype (Lit x)      = Lit x
> > > untype (Succ t)     = Succ (untype t)
> > > untype (IsZero t)   = IsZero ((untype t)::Term Untyped ())
> > > untype (If c t e)   = If ((untype c)::Term Untyped ()) 
> > >		           ((untype t)::Term Untyped ())
> > >		           ((untype e)::Term Untyped ())
> 
> I just checked that the only required change is the one
> in type signature. But I can't explain it at this moment.

I tried to implement another function:

  mapChildren :: (forall a. Term f a -> Term f a) -> Term f b -> Term f b
  mapChildren fun t@(Lit x) = t
  mapChildren fun (IsZero t) = IsZero (fun t)
  mapChildren fun (Succ t) = Succ (fun t)
  mapChildren fun (If c t e) = If (fun c) (fun t) (fun e)

It is supposed to be similar to gmapT from Data.Generics - apply 'fun'
to every immediate Expr child of the node. Unfortunately, I don't know
how to make it compile. I tried a couple of different type signatures.

Best regards
Tomasz

-- 
I am searching for a programmer who is good at least in some of
[Haskell, ML, C++, Linux, FreeBSD, math] for work in Warsaw, Poland