Hello, (Malcolm, sorry for the double post, I forgot to CC the list) I was thinking mostly about the "old-time"-y punning, where I can write a label, say "theField", and it automatically gets expanded to "theField = theField", in record patterns and record constructing expressions. The only corner case that I can remember about this is the interaction with qualified names, the issue being what happens if a label in a pun is qualified?  I think that in such cases we should just used the unqualified form for the variable associated with the label.  In patterns, I can't think of any other sensible alternative.  In expressions, I could imaging expanding "A.theField" to "A.theField = A.theField" but it seems that this would almost never be what we want, while in all the uses I've had "A.theField = theField" is what was needed. I think that this is exactly what GHC implements, at least based on the following example: module A where data T = C { f :: Int } {-# LANGUAGE NamedFieldPuns #-} module B where import qualified A testPattern (A.C { A.f }) = f testExpr f                = A.C { A.f } I imagine that this is fairly close to what was in Haskell 1.3?   As far as wild-cards are concerned, I don't feel particularly strongly about them either way (I can see some benefits and some drawbacks) so I'd be happy to leave them for a separate proposal or add them to this one, depending on how the rest of the community feels. -Iavor On Wed, Feb 24, 2010 at 1:35 AM, Malcolm Wallace wrote:

...

I'd like to propose that we add record punning to Haskell 2011.

Can you be more specific?  Do you propose to re-instate punning exactly as it was specified in Haskell 1.3?  Or do you propose in addition some of the newer extensions that have been recently implemented in ghc (but not other compilers), such as record wildcards?

Regards,    Malcolm

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