On 06/09/06, David Roundy <droundy@xxxxxxxxx> wrote:
Before something like this were to be implemented, I'd rather see
automatic strictifying for avoidence of stack overflows (i.e. when the
stack starts filling up, start evaluating things strictly), which is
most likely another insanely dificult bit of compiler code, which
would also involve figuring out which code paths would save memory,
and which would increase the memory use.
Keep in mind that strict and lazy semantics aren't interchangeable,
that is automatic strictifying could change program behaviour unless
the compiler knew which bits need to be lazy. If it knew that, we
might as well adopt the 'strict where possible, lazy when necessary'
-David House, dmhouse@xxxxxxxxx
Haskell-Cafe mailing list