Hi Cam, Please submit the modified version, and, if you want, create a separate ticket for "seqable?". I would like to review the latter separately.
Stu > Hi again, I modified my-flatten to return the empty list for sets and > maps as core/flatten does. It doesn't seem to handle Arrays anymore > though. I'm assuming it's because ArrayList and (int-array ...) don't > implement Sequential. None the less should I still submit this > modified version that behaves just like core/flatten? > > (defn my-flatten > [coll] > (lazy-seq > (when-let [coll (if (sequential? coll) (seq coll))] > (let [x (first coll)] > (if (sequential? x) > (concat (my-flatten x) (my-flatten (next coll))) > (cons x (my-flatten (next coll)))))))) > > Might it be worth promoting "seqable?" to core? In that case flatten > would handle pretty much everything you could throw at it like you'd > expect. I don't speak for everyone but when I saw sequential? I > assumed it would have the semantics that seqable? does. > > > On Jul 13, 11:04 am, Cam <dlocpuw...@gmail.com> wrote: >> Hi Stuart, >> >> Thanks for checking that out for me! Sorry for not realizing in the >> first place. >> >> I of course would be happy to submit a patch. Should I submit that >> here or over on the assembla page? >> >> On Jul 13, 9:10 am, Stuart Halloway <stuart.hallo...@gmail.com> wrote: >> >> >> >>> Hi Cam, >> >>> Your tests aren't testing the interesting part without a doall. >> >>> That said, my quick tests with doall show your approach faring even better. >>> :-) Also, I think what my-flatten does with Java arrays is intuitive (and >>> the current flatten not so much). >> >>> A patch that preserves the semantics of the existing flatten (except for >>> working with Java arrays) would be welcome. >> >>> Thanks! >>> Stu >> >>>> Another flatten thread! Sorry.. >> >>>> Hello all, before I realized there was a flatten in the master branch >>>> (and before I looked at contrib) I wrote this pretty standard code: >> >>>> (defn my-flatten [coll] >>>> (lazy-seq >>>> (when-let [coll (seq coll)] >>>> (let [x (first coll)] >>>> (if (sequential? x) >>>> (concat (my-flatten x) (my-flatten (next coll))) >>>> (cons x (my-flatten (next coll)))))))) >> >>>> (There's very similar versions on the boards. I'm not claiming this is >>>> anything amazing or unique.) >> >>>> It's not as elegant as what's in core, but in my micro benchmarks (ran >>>> on my laptop; 2.26 core 2 and 4gb ram) it seems to perform a bit >>>> better, _especially_ in the already flattened case. It behaves just >>>> like core/flatten except that it doesn't return an empty list when >>>> passed a map or set, it just returns whatever you gave it but with the >>>> top level converted to a seq. I'm pretty much a clojure noob, so are >>>> there any hidden detractors of this implementation as opposed to the >>>> version introduced in 1.2? >> >>>> Also, quick note, if you swap the call to sequential? with seqable? >>>> from contrib/core, it flattens maps and sets like you'd expect as >>>> well. >>>> Here is how it looks >>>> user=> (my-flatten #{1 {2 3} 4 [5 6 7 #{8 {9 10}}]}) >>>> (1 2 3 4 5 6 7 9 10 8) >> >>>> And for the micro-benchmarks (using "sequential?"): >> >>>> user=> (time (dotimes [_ 1e7] (flatten [1 2 3 4]))) >>>> "Elapsed time: 14,661.592 msecs" >>>> nil >> >>>> user=> (time (dotimes [_ 1e7] (my-flatten [1 2 3 4]))) >>>> "Elapsed time: 922.268 msecs" >>>> nil >> >>>> user=> (time (dotimes [_ 1e7] (flatten [1 [2 [3 [4 [5 [6 [7 [8] >>>> [[[9]]] 10 [11] 12 [13 14 [15]]]]]]]]]))) >>>> "Elapsed time: 18,147.959 msecs" >>>> nil >> >>>> user=> (time (dotimes [_ 1e7] (my-flatten [1 [2 [3 [4 [5 [6 [7 [8] >>>> [[[9]]] 10 [11] 12 [13 14 [15]]]]]]]]]))) >>>> "Elapsed time: 6,088.914 msecs" >>>> nil >> >>>> user=> (time (dotimes [_ 1e7] (flatten [[1 2 3 4 5 6 7 8 9 10]]))) >>>> "Elapsed time: 11,696.693 msecs" >>>> nil >> >>>> user=> (time (dotimes [_ 1e7] (my-flatten [[1 2 3 4 5 6 7 8 9 10]]))) >>>> "Elapsed time: 1,533.983 msecs" >>>> nil >> >>>> Thoughts? >> >>>> -- >>>> 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 > > -- > 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 -- 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