On 11/10/10 4:59 PM, Dan Doel wrote:
I'll admit, the Agda overloading is handy. But I've always considered
Haskell's lack of ad-hoc overloading to be a feature. Type classes give
sensible types for what would normally be ad-hoc. Adding back ad-hoc functions
that have no available general type feels like a step backward. Perhaps that's
just me, though.
It's not just you.
I'd rather see a complete rewrite of the record system in order to make
it into a real record system (perhaps by partly unifying records and
modules, a la Agda and many other dependently typed languages). And I
think TDNR is an abomination that merely paints over the problem; it
doesn't solve the actual issue, and it only makes it harder to implement
a solution later.
--
Live well,
~wren
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
