Path: utzoo!utgpu!watserv1!watmath!att!hriso!attdso!westmark!mole-end!mat
From: mat@mole-end.UUCP (Mark A Terribile)
Newsgroups: comp.lang.c++
Subject: Re: const
Summary: overload on the const
Message-ID: <436@mole-end.UUCP>
Date: 8 Sep 90 04:50:50 GMT
References: <10863@cadillac.CAD.MCC.COM>
Organization: mole-end--private system.  admin: mole-end!newtnews
Lines: 44

In article <10863@cadillac.CAD.MCC.COM>, vaughan@puma.cad.mcc.com (Paul Vaughan) writes:
 . . .
> So, here's a const question.
 . . .
> Consider the following function
  . . .
> const char* substring(const char* string, const char* sub) 
> // substring searches for the first occurence of sub in string and
> //returns a pointer to it or 0 if none is found.
. . .
> [But what if I want to operate on non-const characters??]

Yup, you've got a real problem, though not an insurmountable one.

Starting with Release 2.0, you should be able to overload on
const/non-const-ness .  You can't do it on the basis of the return
type, but you can do it on the basis of the first string.

	const char* substring( const char* string, const char* sub );
	char* substring( char* string, const char* sub );

Does this mean that you have to implement it twice (ignoring that the first
substring() may already exist elsewheres)?  Weeeeeelllll ...

I don't think it unreasonable to write

	inline char*
	substring( char* string, const char* sub )
	{
		return (char*) substring( (const char*) string, sub );
	}

There are some places where this sort of thing won't work; you really need
different algorithms.

It would be nice to be able to say that a function's return type is const
if certain actual arguments are const.  This is probably not something we
will see soon, except perhaps in templates.  Unfortunately, when you use
templates, it is expected (last time I checked) that a weaker form of
argument matching will be applied than for ordinary functions.
-- 

 (This man's opinions are his own.)
 From mole-end				Mark Terribile