Re: Re: template haskell (reify bug and existentials)
Ian Lynagh <[email protected]>
| Newsgroups | gmane.comp.lang.haskell.template |
|---|---|
| Message-ID | <[email protected]> |
[Existentials should work in the CVS head now] On Tue, Nov 02, 2004 at 12:18:51PM +0000, Keean Schupke wrote: > Ian, What are the chances of getting fundeps added to class declarations? > I only need to generate them so support in the syntax and pretty-printer > is all that is required... Here are the changes I have made to my local copy > to get it working. I suppose this breaks existing code! Perhaps you would > rather have a new constructor for classes with fundeps? We have so far taken the attitude that it is better to get a cleaner interface than to protect old code, so I think the same principle should apply here. > in Syntax.hs: > ... > type Fundep = ([Name],[Name]) -- add to export list Experience has suggested that data Fundep = FunDep [Name] [Name] is more convenient as it allows you to make class instances for it. > | ClassD Cxt Name [Name] [Dec] [Fundep] I think ClassD Cxt Name [Name] [Fundep] [Dec] is nicer as it matches the concrete syntax order. Has anyone got any objections to this (in particular the breaking existing code aspect)? Thanks Ian