[Template-haskell] RE: Question regarding template Haskell

ozone at algorithm.com.au ozone at algorithm.com.au
Thu Sep 18 11:15:59 EDT 2003

On 17/09/2003, at 9:07 PM, Simon Peyton-Jones wrote:

> | If your 'x' variable is monomorphic, you can use the rType function
> | that Sean and I came up with, detailed here:
> |
> |  
> http://www.haskell.org/pipermail/template-haskell/2003-July/ 
> 000128.html
> |
> | That currently handles only 2-tuples, but it shouldn't be too hard to
> | extend it to work on any-size tuples.  Let me know if you're having
> | problems.
> I missed that message.  How does rType differ from typeOf, defined in
> Data.Typeable?

typeOf :: a -> TypeRep

rType :: a -> Typ (or :: a -> Type in the new THSyntax definitions).

The two are conceptually the same, but they return you different data  
types: typeOf returns you the TypeRep which the Data.Dynamic functions  
use, and rType returns you a Typ (now Type) which THSyntax uses.   
Another way to think about it is that rType is  Template Haskell's  
reifyType as a function.

The point is that you can use rType to return you the definition of a  
Type given in THSyntax, which you can then use directly within  

     *Mocha.ReifyType> Data.Dynamic.typeOf (undefined :: (Int, Char))
     *Mocha.ReifyType> rType (undefined :: (Int, Char))
     AppT (AppT (TupleT 2) (ConT "GHC.Base:Int")) (ConT "GHC.Base:Char")

Come to think of it, now that we have a concrete syntax, perhaps it'd  
be worth merging the definitions of THSyntax's Type and Data.Dynamic's  
TypeRep?  They're conceptually the same thing, but with different data  
representations ...

A slightly wilder idea (which is not a serious suggestion) would be to  
use Template Haskell to implement Dynamics.  If there's a pre-defined  
"Dynamic" data type, all the various toDyn and fromDyn functions could  
be meta-programmed so that they emit code to convert to and from the  
Dynamic type.  Effectively you could use Template Haskell as a  
generics/polytopic system.  I have no idea whether this is possible,  
it's just a fleeting thought.

BTW, you didn't miss that message Simon -- you even replied[1] to it :).


% Andre Pang : trust.in.love.to.save

More information about the template-haskell mailing list