2010/12/29 Ken Wesson <kwess...@gmail.com>
> On Wed, Dec 29, 2010 at 11:55 AM, Laurent PETIT <laurent.pe...@gmail.com>
> wrote:
> >
> >
> > 2010/12/29 Ken Wesson <kwess...@gmail.com>
> >>
> >> On Wed, Dec 29, 2010 at 11:42 AM, Laurent PETIT <
> laurent.pe...@gmail.com>
> >> wrote:
> >> > 2010/12/29 Ken Wesson <kwess...@gmail.com>
> >> >> (defn insert-into [s x]
> >> >> (let [[low high] (split-with #(< % x) s)]
> >> >> (concat low [x] high)))
> >> >>
> >> >> (defn insertion-sort [s]
> >> >> (reduce insert-into [] s))
> >> >
> >> > Hello, just a little 0.00002€ : insert-into will return seqs, so the
> >> > reduce
> >> > could read :
> >> > (reduce insert-into () s)
> >> > to make it clear that it's seqs end to end inside insertion-sort
> >>
> >> It's my habit to use a vector for any "seq literal".
> >>
> >> How about I split the difference?
> >>
> >> (reduce insert-into nil s)
> >>
> >> seems to work just as well. :)
> >
> > Yeah, not a big deal, it's just that by just reading the line (reduce
> > insert-into [] s), I see the initial "collected value" is a vector, and I
> > make an assumption about insert-into to return vectors as well.
>
> Eh? I don't. I tend to only assume "it takes seqables" if I see vector
> literals going into something. If it's a reduction, I presume the
> function being reduced to return seqables as well.
>
> My preference here is for the empty vector:
>
> * The empty list () is odd. In Clojure usually parentheses wrap an
> executable expression, not just data.
> * Quoting it -- '() -- is just plain ugly.
> * The value nil can generally stand in for an empty seq, but has
> other unrelated meanings as well.
> * The empty vector is, explicitly, a zero-length seqable piece of
> inert data.
> * So is the empty set #{}, but it's commonplace to use vector
> literals as "seq literals", so an empty vector is less confusing.
> My objection to the empty set is the same as yours to the empty
> vector.
> * That goes double for the empty map {}.
>
To the risk of repeating myself and not having totally understood your above
explanation, worded differently :
when I see (reduce insert-into [] s) with [] in the 'val' position, I use
this as a hint on what the resulting val returned by reduce will be.
If we're ok with the above reasoning, then I find that [] implies the result
will be a datastructure (and a vector datastructure, to be even more precise
IMHO).
While I see the () value as more "neutral" (though not ideal as you said,
but still better than [] in the particular case).
--
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
