
23 Jan
2007
23 Jan
'07
3:50 p.m.
Hi
prove/compute anything you couldn't before. While removing _|_ from the language does make some things nicer to reason about, there aren't many corners where _|_ really gets in the way that much - seq being one of those few corners.
But that is exactly the problem: `seq` forces _|_ to get into the way, when it normally doesn't. So I'm not clear that trying to fit `seq` into a formalization of Haskell's semantics is the way to go.
Agreed, that was the point I was trying to make :) You seemed to be suggesting _|_ was "evil" (for want of a more precise term) in its behaviour with Haskell. As you seem to say now (and I agree), _|_ is a perfectly useful value, just seq gets in the way. Thanks Neil