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
