UT Lisp Users Group?

Brad Dorner bdorner at novell.com
Thu Jul 14 09:27:45 MDT 2005


On Wed, 2005-07-13 at 14:40 -0700, Jonathan Ellis wrote:
> 
> It's been well-demonstrated that some languages enable greater
> productivity than others.  For example, "An empirical comparison of C,
> C++, Java, Perl, Python, Rexx, and Tcl"
>
(http://page.mi.fu-berlin.de/~prechelt/Biblio/jccpprt_computer2000.pdf).
>  See in particular the "total time for programming [hours]" graph on
> page 6.
> 
> Dismissing the almost self-evident assertion that all languages are
not
> equal as "snobbery" baffles me.  Sure, you can bring out the
irrelevant
> "as long as they are both turing complete, who cares" but here in the
> real world, some languages are clearly better than others.
> 
> Sendmail configuration is turing complete, after all. :) 
> (http://okmij.org/ftp/Computation/sendmail-as-turing-machine.txt)
> 
> -Jonathan

Please note that in the study referred to above it says 

"...The programs analyzed in this report come from two different
sources. The Java, C, and C++ programs were produced in 1997/1998 during
a controlled experiment..."

First not that the above statement from the document state that the
information for the programming languages was from programs written in
1997 and 1998. The Java language was only around for 2 years and was
just starting to get a following at that time. There were no great Java
programmers at that time. C and C++ Had a larger following and had been
around for some time. The languages were also much more refined at that
time. 

The following line also appears in the document that you referred to.
The statement shows only a bias as opposed to proof, sense there is no
supporting evidence to back it up.


"...In the script group, the Perl subjects may be more capable than the
others, because the Perl language appears more than others to attract
especially capable people..."

Now for the disclaimer in the document. The following document segment
says the results are only valid - if valid can be used for this "study"
- for the phonecode problem.

"...However, it must be emphasized that the results are valid for the
phonecode problem only; generalizing to different application domains
would be haphazard..."

PLEASE - if you are going to use words like 'well-demonstrated' make
sure that it is actually demonstrated. The comparison that you
referenced isn't even a good example of a comparison.

Please note also that the document is more the 5 years old. In the
computer world that would make it invalid in and of itself. Try and get
a college to accept a computer class that you took over 5 years ago for
credit. I've had to change programming languages every 3 years for the
past 20+ years. 

Now for my 2 cents.

There is no good programming language, no better nor worst languages.
Program languages are written to meet specific needs and therefore all
languages are good for what they are meant to do. 

Programming is also an art form, No two people will program the same or
even choose the same tools to complete the same task. 

sorry I am getting long winded. 

No such thing as the perfect or best language.

Brad









More information about the PLUG mailing list