GHC already ignores the existence of seq, for instance when doing list fusion.
The unconstrained seq function just ruins too many things.
I say, move seq top a type class (where is used to be), and add an unsafeSeq
for people who want to play dangerously.

  -- Lennart

On Tue, May 6, 2008 at 3:27 PM, Luke Palmer <lrpalmer@gmail.com> wrote:
On Tue, May 6, 2008 at 2:50 AM, apfelmus <apfelmus@quantentunnel.de> wrote:
>  Concerning the folklore that  seq  destroys the monad laws, I would like
>  to remark that as long as we don't apply  seq  to arguments that are
>  functions, everything is fine. When  seq  is applied to functions,
>  already simple laws like
>
>   f . id = f
>
>  are trashed, so it's hardly surprising that the monad laws are broken
>  willy-nilly. That's because  seq  can be used to distinguish between
>
>   _|_ :: A -> B    and   \x -> _|_ :: A -> B
>
>  although there shouldn't be a semantic difference between them.

It seems that there is a culture developing where people intentionally
ignore the existence of seq when reasoning about Haskell.  Indeed I've
heard many people argue that it shouldn't be in the language as it is
now, that instead it should be a typeclass.

I wonder if it's possible for the compiler to do more aggressive
optimizations if it, too, ignored the existence of seq.  Would it make
it easier to do various sorts of lambda lifting, and would it make
strictness analysis easier?

Luke
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe