Path: utzoo!utgpu!water!watmath!clyde!cbosgd!ihnp4!chinet!nucsrl!gore
From: gore@nucsrl.UUCP (Jacob Gore)
Newsgroups: comp.lang.c++
Subject: Re: Operator overloading considered harmful
Message-ID: <8180001@nucsrl.UUCP>
Date: 10 Jan 88 21:28:21 GMT
References: <240@vsi1.UUCP>
Organization: Northwestern U, Evanston IL, USA
Lines: 36

/ nucsrl:comp.lang.c++ / steve@vsi1.UUCP (Steve Maurer) / Jan  8, 1988 /
>    I have a bone to pick with operator overloading. [...]
>Consider the following:
>
>    foo(int i)
>    {
>    i #= i ^^ i @& i;
>    }
>
>what does it do??  [...]

It was my impression that in C++ you could not define operators with new names
-- only overload existing names.  (The overloaded operators keep the
associativity and precedence qualities of the built-in operators of the same
name, right?)

But anyway, no language can prevent a determined nut from writing programs
with confusing semantics.  What should we "consider harmful" next,
alphanumeric variable names?  Consider the following:

	nonobservance(int depressant)
	{
	hoke = unhampered + serf / chiaroscuro;
	}

What does IT do?

If a programmer does wants to write clear programs, then a better language
will support that goal, while a worse language will get in the way.  Names of
operators should be picked in such a way that they make sense for the context
in which they are used.  This is harder to do when you cannot define your own
operators, and harder yet to do when you cannot even overload predefined
operator names. 

Jacob Gore				gore@EECS.NWU.Edu
Northwestern Univ., EECS Dept.		{gargoyle,ihnp4,chinet}!nucsrl!gore