Is this a dream or a nightmare? (Was Re: XML)

[Alex Martelli]

"Emile van Sebille" <emile at fenx.com> wrote in message
news:8rq496$ik0l1$1 at ID-11957.news.cis.dfn.de...
    [snip]
> One recent study that many of this group participated in
> shows that python uses less lines of code than the other
> languages.
>
> http://wwwipd.ira.uka.de/~prechelt/Biblio/jccpprtTR.pdf

Excellent reference, but please note that the median length
(in SLOC -- "non-comment, non-blank source lines of code")
was (by a tiny and not statistically significant fraction) even
shorter in Perl than in Python -- maybe by about as 5% (which
makes one smile thinking of those claimed "25 times", but...).
About 2/3 times shorter than median Java/C/C++ programs.

Median time for task completion (around 2.5 hours) is also
statistically indistinguishable between Perl and Python (the
tiny non-significant advantage is here in Python's favour), and
about 4 times better than Java/C/C++ programs.

Productivity in SLOC/hours was where Python really stood
out from all other languages, around 40 versus a range of
20 to 30 for the other languages.

But basically the task turned out into "scripting (Python, Perl,
Tcl, Rexx) versus traditional compiled languages (C, C++,
Java)", and a significant victory for scripting (4 times better
productivity, substantially-equal run times -- maybe 30%
slower than C or C++, but faster than Java...).  Pity the author
did not think (or was unable to) include functional programming
languages (ML dialects, Haskell, Erlang, perhaps Dylan...) --
the productivity/speed performances of such languages in the
yearly ICFP competition makes it appeat they might well have
shown up as significant contenders too!


Still, a very interesting study indeed -- wish there were more
studies like it...!

Alex