(#~a:~:{:"1)(]/.~/:~&>)<;._2]1!:1<'t'
In Henry's solution:
- missing the ] argument to /.
- can use /:~&> instead of /:&.> (will be faster, I expect)
- cutting into words can be done more succinctly
- using a shorter file name will make soln shorter
----- Original Message -----
From: Henry Rich <[EMAIL PROTECTED]>
Date: Thursday, September 25, 2008 12:16
Subject: RE: [Jprogramming] Words of Equal Characters
To: 'Programming forum' <[email protected]>
> I don't have the file to test with, but I offer
>
> (#~a:~:{:"1)(/.~/:~&.>)<;._2 toJ fread
> jpath'~temp\unixdict.txt'
>
> Use the sorted strings as the keys without converting to
> numbers; don't box the results but allow them to fill
> a table; longest sets will be the ones without a: at the end.
>
> Henry Rich
>
> > -----Original Message-----
> > From: [EMAIL PROTECTED]
> > [mailto:[EMAIL PROTECTED] On Behalf Of Dan Bron
> > Sent: Thursday, September 25, 2008 2:55 PM
> > To: Programming forum
> > Subject: [Jprogramming] Words of Equal Characters
> >
> > Recently, I've been playing around on Rosetta Code [1].
> A
> > Python guy posted a new task:
> >
> >
> http://rosettacode.org/wiki/Words_Of_Equal_Characters>
> > What is the shortest possible J expression to solve this
> > task? Here, "shortest" means the minimum
> number of
> > essential (required) characters. Right now, I have:
> >
> > >(#~[:(=>./)#&>)(</.~[:i.~/:~&.>)<;._2 toJ fread
> > jpath'~temp\unixdict.txt'
> > +-----+-----+-----+-----+-----+
> > |abel |able |bale |bela |elba |
> > +-----+-----+-----+-----+-----+
> > |alger|glare|lager|large|regal|
> > +-----+-----+-----+-----+-----+
> > |angel|angle|galen|glean|lange|
> > +-----+-----+-----+-----+-----+
> > |caret|carte|cater|crate|trace|
> > +-----+-----+-----+-----+-----+
> > |elan |lane |lean |lena |neal |
> > +-----+-----+-----+-----+-----+
> > |evil |levi |live |veil |vile |
> > +-----+-----+-----+-----+-----+
> >
> > which is shorter than the original Python solution, but not
> > dramatically shorter. Is there a way to code this to
> really
> > show off J's power? I'm more focused on the aspect that
> > produces the result given a list of words, rather than the
> > aspect that generates the list of words from the text (i.e.
> > I'm seeking to minimize the code to the left of the cut).
> >
> > -Dan
> >
> > [1] Rosetta Code is what it sounds like -- a chrestomathy
> > site where programmers are invited to solve programming tasks
> > using different languages, to build up a "Rosetta Stone" for
> > different programming languages.
> >
> > Of course, the primary motivation for posting a solution to
> > Rosetta Code is not to teach others your language, or to make
> > it easier for a maintenance programmer to convert a legacy
> > system from COBOL to Perl. Those are just side
> effects.
> > The primary motivation for posting a solution to Rosetta Code
> > is to brag. It's main constituency is evangelists.
> It's the
> > premier public programming pissing match.
> >
> > One drawback of Rosetta Code is that only the results of the
> > programmer's efforts are visible; there's no way to tell how
> > long the solution took, or how the language enabled the
> > programmer to reach the solution. So even though this
> > solution (the "obvious way") took me less than a minute to
> > write, no one reading the page will know that. The only
> > opportunity to "wow" the audience is with a succinct, elegant
> > solution. And since most of the audience won't be able
> to
> > read J, elegant is less important than succinct. That's
> why
> > I'm seeking an impressively short solution to this task
> > (which seems right up J's alley).
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
