Newsgroups: comp.lang.misc
Path: utzoo!utgpu!news-server.csri.toronto.edu!helios.physics.utoronto.ca!alchemy.chem.utoronto.ca!mroussel
From: mroussel@alchemy.chem.utoronto.ca (Marc Roussel)
Subject: Re: Multi-compilers
Message-ID: <1990Sep24.210407.7714@alchemy.chem.utoronto.ca>
Keywords: design, source
Organization: Department of Chemistry, University of Toronto
References: <2576@l.cc.purdue.edu> <9206@ccncsu.ColoState.EDU> <2581@l.cc.purdue.edu>
Date: Mon, 24 Sep 90 21:04:07 GMT

In article <2581@l.cc.purdue.edu> cik@l.cc.purdue.edu (Herman Rubin) writes:
>Why is there an absolute value function rather than an absolute value
>operator?  Only because when Fortran was written, they were limited to
>a 48 character set (IBM punched cards, at the time).  This is also why
>** was used for exponentiation, and even * for multiplication.  I have
>used a dialect of Fortran which overloaded some of the "standard" functions,
>so that absolute value, logarithm, mod, etc., did not have to have any modifier
>indicating the type(s) of the arguments.  The compiler did the appropriate
>analysis.

Uhhh... Isn't this just what the FORTRAN 77 standard dictates?  (Yes, I
know about the exceptions.  I mean in the usual case where I say, for
instance, a = sin(b) and the types of both a and b are well-defined.)

                                Just asking...

				Marc R. Roussel
                                mroussel@alchemy.chem.utoronto.ca