Subject: Re: Theory #51 (superior(?) programming languages)
From: Erik Naggum <erik@naggum.no>
Date: 1997/01/26
Newsgroups: comp.arch,comp.lang.lisp,comp.lang.scheme
Message-ID: <3063274536533203@naggum.no>


* Scott Schwartz
| My point is valid: you want to treat lisp like a shell, but then ignore
| the cost of doing so when the overhead is in question.  In contrast, the
| time you gave for the a.out version includes the cost of dynamically
| loading all the shared libraries.

your "point" is not only invalid, it is completely ridiculous.

I measured 1 million calls to these functions, and got the timings I did,
and don't even try to believe I fork'ed and exec'ed a million times.  I
wanted to measure the time of the function itself, not the time of running
a program.  by running the same function a million times inside a given
process, the time information I get for Lisp can be expressed as

    1000000(binomial+looper)

and for the C as

    1000000(binomial+looper)+startup

to find the time of the looper and the startup costs, I ran both with
(binomial 35 35) to find the cost of the function call.  those timings were
subtracted from the times I listed.

considering the amount of idiocy I get from C users when timing stuff and C
loses or proves to be less than the super-efficient hands-down winner of
all contests, my belief that C damages people's brains in irreparable ways
is indeed solidifying.

#\Erik
-- 
1,3,7-trimethylxanthine -- a basic ingredient in quality software.