Perf. Python/Tcl comparison ?

[Cameron Laird]

In article <37DD5B8B.9402908F at westgroup.com>,
Elie Naulleau  <Elie.Naulleau at westgroup.com> wrote:
>
>I wondering if some specs have been done about comparing
>Tcl and Python performances, on strings integers and floats for
>instance.
>Thks,
>
>Elie.Naulleau at westgroup.com
>

Yes.

This is the best summary:  consider their performances
indistinguishable.  Your organization would be a very
unusual one for the differences in performance between
the two languages to matter *at all* in comparison to
the profounder differences the two exhibit:  programming
paradigm; importance of COM facilities; integration with
allied key applications (Zope, Tk, Expect, PyMath, ...);
Unicode-awareness; ...

Second key fact:  if performance measurement is
exceptionally determinative for you, you will find it
more rewarding to construct your own benchmarks than to
rely on the testimony of others.  I write this not from
general ideological nihilism about benchmarking; to the
contrary, I regularly salute good work done in this
area.  However, the costs of writing your own test
programs in these two languages are small.  Moreover,
the applicability of others' work on performance with
these two languages is a matter for expert judgment,
because both are evolving so rapidly.  Measurements
made in 1997 might be two (!) orders of magnitude dif-
ferent from identical configurations done today.

If you insist on pushing beyond these two propositions,
which should structure most of your thinking, yes, a
few generalizations are possible.  Heavily arithmetic
code will probably perform better in Pythonian source.

I repeat:  you'll be better off thinking about other
ways Tcl and Python differ.
-- 

Cameron Laird           http://starbase.neosoft.com/~claird/home.html
claird at NeoSoft.com      +1 281 996 8546 FAX