On 05 October 2005 17:05, Lennart Augustsson wrote:

Wolfgang Jeltsch wrote:

...

Am Mittwoch, 5. Oktober 2005 16:22 schrieb Simon Marlow:

...

[...]

...

Also, GHC's optimiser currently treats (_|_ :: IO a) and (do _|_; return ()) as interchangeable, which is naughty, and people have occasionally noticed, but the benefits can sometimes be huge. It is this distinction that makes it hard to optimise IO code in a Haskell compiler, though.

I think, seq should be a method of a type class. Then we could forbid applying seq to a function, we could forbid applying seq to an IO expression and we could forbid applying seq to expressions of any type with hidden implementation for which we don't want to provide bottom tests.

I agree with you. And that is how it used to be, but then some people didn't think that was convenient enough so now we are stuck with a seq that (IMHO) stinks. :)

Having a seq that works on anything is occasionally very useful for fixing space leaks, and the type class version was kind of ugly and unwieldy. The type class pops up when you use ! in a data declaration, which looks strange. But I agree that polymorphic seq does have some painful side effects. Cheers, Simon