begin quoting Andrew Lentvorski as of Sat, Jan 12, 2008 at 04:03:26PM -0800:
> SJS wrote:
>
> >Finding a good problem is a wonderful thing.
> >
> >A suduko solver seems like a good fit for a lispy language. :)
>
> I was less interested in the "learning the language" issue than
> Stewart's "do it all without a reference".
>
> To my mind, the second task was actually a good test of the language.
Ah!
Yes, I see. Especially a language that one hasn't used in awhile; had
I been using perl heavily in the previous week, it wouldn't have been
hard in perl... as I hadn't "used perl in anger" for quite some time,
it would be hopeless to attempt that program in perl.
It wouldn't have been any trouble in Java, but that's because I use
Java all the time. Wouldn't be a good test.
> The fact that Python was good enough that I remembered 98% of it, had to
> test about 1%, and had to run grep on the installed Python codebase for
> the last 1% ("argv" is in "sys" ... ah, right) is a good stress test.
Non-trivial but simple program in a language you used extensively but
haven't touched in awhile, without significant reliance on references.
Hm. Say I were to do this in Smalltalk (gc, lots of built-in datatypes);
would it be fair (in terms of a 'stress test') to use the GUI development
environment?
> Besides, I'm not interested in problems that fit a language well. I'm
> interested in problems that *don't* fit the language well. That's the
> test. You don't always know the form of the solution to a problem a
> priori.
You can write a compiler in COBOL. Doesn't mean that one /should/; but
if it's the best tool around, it's better than nothing.
--
I should perhaps collect the comments and bugs
And fix the code up and polish it for you lugs.
Stewart Stremler
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-lpsg
