Dave writes:
Why doesn't Haskell allow you to name components? I know that you
don't *need* to name them, but, like Sandra, I have also seen data
structures with almost two dozen fields. Pattern matching is nice,
but it seems like changing the representation of something could
potentially require a lot of code changes.
I would guess a lack of energy on the part of the committee. This
doesn't involve any deep theory but there are a bunch of engineering
issues. I think everyone would agree that this would be useful. Of
course you could extend this a bit and decide to add records to the
type system and then things get really hairy. This also is similar
to some sort of `view'-like extension to pattern matching - another
controversial topic on the committee.
John
