Hi, Disclaimer: believing this kind of benchmark. No clue whether the following is really valid.
On 5 Apr., 09:44, Ken Wesson <kwess...@gmail.com> wrote: > If I understand your claim correctly, it's therefore wrong. Since it's correct, you don't understand my claim correctly. The numbers in between are the memrem runs. Using your partition-by. user=> (do ...) ; long command snipped ----> Start empty. ----> (def s nil) 0 1 2 3 4 5 6 ----> (def s (partition-by alength (map byte-array [100000000 100000000 100000000 200000000]))) 0 1 2 3 4 5 6 ----> Nothing realised, yet. Get first group. ----> (first s) 0 1 ----> Mem used. Drop head, get rest. ----> (def s (rest s)) 0 1 ----> Head is hold? Realise rest. ----> (def s (seq s)) 0 1 2 3 4 ----> References gone. Memory freed. nil Using core partition-by: user=> (do ...) ; long command snipped ----> Start empty. ----> (def s nil) 0 1 2 3 4 5 6 ----> (def s (clojure.core/partition-by alength (map byte-array [100000000 100000000 100000000 200000000]))) 0 1 2 3 4 5 6 ----> Nothing realised, yet. Get first group. ----> (first s) 0 1 ----> Mem used. Drop head, get rest. ----> (def s (rest s)) 0 1 ----> Head is hold? Realise rest. ----> (def s (seq s)) 0 1 ----> Huh? nil This was surprising and turned out to be a bug in core partition-by. Using fixed core partition-by: user=> (do ...) ; long command snipped ----> Start empty. ----> (def s nil) 0 1 2 3 4 5 6 ----> (def s (fixed-core-partition-by alength (map byte-array [100000000 100000000 100000000 200000000]))) 0 1 2 3 4 5 6 ----> Nothing realised, yet. Get first group. ----> (first s) 0 1 ----> Mem used. Drop head, get rest. ----> (def s (rest s)) 0 1 2 3 4 ----> Note that head is *not* hold, references to first are dropped. nil Interesting where such discussions can lead: it helped to uncover an issue on an edge case with core partition-by. Sincerely Meikel -- 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
