[Haskell-cafe] Re: Efficient or predictable Ord for Typeable

[Ian.Stark at ed.ac.uk]

On Tue, 30 Nov 2004, Lennart Augustsson wrote:
> George Russell wrote:
>> I suggest you implement
>> 
>>    hashTypeable :: Typeable -> IO Int32
>
> And/or
>    mkHashTypeable :: IO (Typeable -> Int32)

And/or (you'll hate this):

   mkHashTypeable :: ACIO (Typeable -> Int32)

After all, I'm sure you don't want hasTypeable / mkHashTypeable to give 
different answers when used twice on the same run.

Simon Peyton-Jones wrote:
> The trouble is that *any* "function" can now deliver unpredictable
> results.  Can I rely on the fact that
>     foo :: Int -> Int
> will always give the same answer given the same input.  Not any more.

Performing a dubious toplevel 'declare hashTypeable <- mkHashTypeable' 
hits an odd middle ground here: you can indeed rely on the fact that

   foo :: Int -> Int

will always give the same answer given the same input, anywhere in the 
program.  It just won't necessarily give the same answer next time you run 
the program.  No call-by-name rewrites are broken, but something is 
certainly strange.

-- 
Ian Stark                               http://www.ed.ac.uk/~stark
LFCS, School of Informatics, The University of Edinburgh, Scotland