[Haskell-cafe] Re: Suggested algorithm to control upper bound of space "leaks"

Shelby Moore shelby at coolpage.com
Mon Nov 2 18:00:15 EST 2009

> stimulating reply
> about the convolution of the competition for the shared resources in the
> external state machine.

Apparently the concurrency problem is potentially convolved with external
state in much more convoluted and intertwinded spaghetti:


Multi-core is coming!  Wouldn't it be better if the mainstream scripting
(i.e. composability) language was also pure FP (and the composable modules
were pure FP).

If it costs 4X slower (lose 3/4 of) performance (and I doubt it would be
any where near that bad) on keeping some of the thunks closures around so
that scripting doesn't require masters degree in profiling, but we can
remain in pure FP and minimize our STM usage (which also purportedly costs
4X), then if we hit 8+ cores then perfomance has improved, not to mention
all the other benefits.  Okay I realize that napkin calculation can't
possibly encompass the details, but just make a conceptual point.

More information about the Haskell-Cafe mailing list