RE: TypeFamilies vs. FunctionalDependencies type-level recursion

[Simon Peyton-Jones]

I have expanded the draft spec on
http://hackage.haskell.org/trac/ghc/wiki/NewAxioms
to answer some of the questions on AntC's discussion page.


From: haskell-prime-boun...@haskell.org 
[mailto:haskell-prime-boun...@haskell.org] On Behalf Of José Pedro Magalhães
Sent: 29 May 2012 11:14
To: AntC
Cc: haskell-prime@haskell.org
Subject: Re: TypeFamilies vs. FunctionalDependencies  type-level recursion

Hi,
On Tue, May 29, 2012 at 11:03 AM, AntC 
anthony_clay...@clear.net.nzmailto:anthony_clay...@clear.net.nz wrote:
Simon Peyton-Jones simonpj@...mailto:simonpj@... writes:


 See also http://hackage.haskell.org/trac/ghc/wiki/NewAxioms
 (as yet unimplemented)

 Simon

Thank you Simon (and Pedro).

Are you inviting comment/suggestions/requests for clarification at this stage?

Definitely! I think the design space should be explored in detail.

There is plenty of prior work/similar ideas to include in the references.

That document is not a paper draft; it's a draft of a design of a new GHC 
extension.

How's the best way to help? (Without unleashing a maelstrom.)

Perhaps having a wiki page where the problem of OverlappingInstances is 
discussed, and alternative solutions are proposed, so that at some point we can 
look at all of them and try to make an informed decision. I think it's good to 
have a wiki page to guide this sort of email discussion.


Cheers,
Pedro


AntC



___
Haskell-prime mailing list
Haskell-prime@haskell.orgmailto:Haskell-prime@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-prime

___
Haskell-prime mailing list
Haskell-prime@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-prime

Re: TypeFamilies vs. FunctionalDependencies type-level recursion

[AntC]

Simon Peyton-Jones simonpj@... writes:

 
 I have expanded the draft spec on
 http://hackage.haskell.org/trac/ghc/wiki/NewAxioms

Thanks Simon, that's much clearer.

By the way, are the examples for the multi- type instance declarations quite 
as intended? The heads have no head, as it were. Did you mean, or is this 
allowed?

type instance F [a] where ...

type instance F (a, b) where ...

(From a documentation point of view, this shows that the instance groups are 
non-overlapping.)

 to answer some of the questions on AntC’s discussion page.
  

(I'd rather you called it just *the* discussion page; I'm doing the ego-less 
contributor thing. I must admit that after I got the page started, I've not 
had so much time to keep building it.)

AntC


___
Haskell-prime mailing list
Haskell-prime@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-prime