One thing that is likely to be surprising is that (.) is
right-associative, whereas in C it's left-associative.
object . action . action would try to parse as object . (action .
action) - which may or may not typecheck but is probably not what was
intended.
On Thu, May 15, 2014 at 11:11 AM, silvio
relevantly: https://ghc.haskell.org/trac/ghc/wiki/Records/DeclaredOverloadedRecordFields...
Ways in which I think my implementation is superior
1) in the proposal the thing behind the dot has to be a function. Therefore, you have to wrap everything in a function before you can use it. Something like this is not possible
[1,2,3] . 0 1
2) I'm not sure this proposal solves the name collision problem but that might just be that i don't understand it.
3) my library doesn't require any additional syntax only a few already existing extensions. And you can use a bit of template haskell to assist the creation of objects.
4) Also the proposal doesn't mention updates but if they have to be a function you cant use the same name as for the function to get a field. And you simply can't beat my syntax, in which you can use the same name for both. object . fieldName := value -- update object . fieldName -- get
Silvio _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe