On Wed, Nov 17, 2010 at 7:55 PM, Robert McIntyre <r...@mit.edu> wrote: > So, just to be clear, > > user> (def nil-seq (doall (interleave (repeat 1e5 nil) (repeat 1e5 > "whatever"))) ) > #'user/nil-seq > > user> (time (doall (keep identity nil-seq))) > "Elapsed time: 122.485848 msecs" > > user> (time (doall (remove nil? nil-seq))) > "Elapsed time: 149.71484 msecs"
I have to run these all a few times before the times quit shrinking (JITting being done). Then: user=> (time (do (doall (keep identity nil-seq)) nil)) "Elapsed time: 70.24324 msecs" nil user=> (time (do (doall (remove nil? nil-seq)) nil)) "Elapsed time: 40.47016 msecs" nil user=> (time (do (doall (filter identity nil-seq)) nil)) "Elapsed time: 36.70256 msecs" nil It seems on my system keep identity is actually almost twice as SLOW as remove nil? and filter identity is slightly faster; filter identity is almost exactly twice as fast as keep identity. Filter identity does, of course, discard falses as well as nils, but when that's acceptable it's worth knowing it's fastest. I wonder if the difference is that I'm using the -server vm? It's especially interesting, in light of the likelihood that identity is probably compiling away to nothing when the JIT is done inlining everything, that both the slowest and the fastest use identity. It suggests that keep is inherently somewhat slow, in particular. -- You received this message because you are subscribed to the Google Groups "Clojure" group. To post to this group, send email to clojure@googlegroups.com Note that posts from new members are moderated - please be patient with your first post. To unsubscribe from this group, send email to clojure+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/clojure?hl=en