Path: utzoo!attcan!uunet!know!cs.utexas.edu!yale!cmcl2!lanl!jlg
From: jlg@lanl.gov (Jim Giles)
Newsgroups: comp.lang.misc
Subject: Re: Fortran vs. C for numerical work - expression notation
Message-ID: <8430@lanl.gov>
Date: 10 Dec 90 20:51:28 GMT
Organization: Los Alamos Natl Lab, Los Alamos, N.M.
Lines: 34

A room full of Fortran programmers may indeed groan when someone
mentions that A > B will be the new form of  relational expression.
They are _not_ groaning because they think it's a bad idea, they
are groaning because it is a good idea that has been discussed to
death in the Fortran community for nearly two decades.  They groan
simply because it is an old debate that they are simply tired of
hearing about.

Now, if you tell that same roomful that you plan to remove x**y and
replace it with pow(x,y), they'll throw vegetables at you.  This is
because such a change is _not_ a good idea.  One of the advantages to
x**y or x^y over pow(x,y) is conciseness (which isn't really a word -
the correct word is 'concision' :-).  Another advantage is that, while
different from standard mathematical notation, they are _MUCH_ closer
to it.  If anyone disagrees, fine - Fortran doesn't FORCE you to use
the '**' operator, you can write a pow() function and call it - you
can even make it a statement function so that it gets inlined.  C
can't offer the same privilege to converting Fortan users - there is
no way in C (within the standard anyway) of defining an exponentiation
operator.  People who find this form most natural (most of us) cannot
use it in C.

Yes, C++ has allows user defined operators (or, allows users to
overload existing operators, but not define new ones is the real
rule).  This is a good idea and _allows_ most of Fortran user's
complaints about C expression notation to be fixed (I say 'allows',
there is still a problem of compatibility if two sets of users pick
different operators to overload, different functions to implement the
overloaded functionality, etc.).  But, if such overloaded operators
are actually implemented as _external_ function calls, this answer
is not satisfactory.  (By the way, Fortran Extended has overloadable
and user definable operators too.)

J. Giles