Re:Ask HN: Multiprocessing & concurrency: C, Python, Clojure...
by fnl on Sunday October 24, 2010 @12:46AM
(#1825229)
Hmm, so you would not say that Clojure's or Go's approaches are worth it in terms of programming ease? The plots you point to looks at least seems worrying: Python is smacked on the concise 0-axis, C on speed. Clojure at least makes no especially good grounds on those plot, as I already expected. However, I was hoping people would say it is so much easier writing a parallel implementation of the code I am using that I switched to Clojure/consider using Go/whatever.
Re:Ask HN: Multiprocessing & concurrency: C, Python, Clojure...
by Detrus on Sunday October 24, 2010 @01:46AM
(#1825344)
The charts are for general language use, not specifically concurrency. Go and Clojure are better than locks and mutexes of C, but there were similar ideas in Scala and other languages along the way and the newcomer improvements don't land far from those.
So Go, Clojure, F#, Scala, Python could proabably be lumped into the same ease of implementation category as far as concurrency. They'll all make a mess if you have complex control flow. This http://news.ycombinator.com/item?id=1824683 (which you've already seen) or http://en.wikibooks.org/wiki/F_Sharp_Programming/Async_Workf... is what I call a mess, but it's as clean as it gets for now.
I think the real test is whether people start using concurrency as the norm and not as a separate use case (since it's annoying to write). I don't think these languages will have that effect.
As far as performance vs implementation here is a handy chart http://shootout.alioth.debian.org/u32/code-used-time-used-sh... That's for generic language things, but concurrency probably doesn't stray far from the overall performance.