NetNews Usenet Archive 1993 #1/NN_1993_1.iso/spool/comp/std/cplus/2012

Text File | 1993-01-08 | 1.8 KB | 40 lines

Newsgroups: comp.std.c++
Path: sparky!uunet!haven.umd.edu!darwin.sura.net!gatech!swrinde!sdd.hp.com!usc!news.service.uci.edu!network.ucsd.edu!qualcom.qualcomm.com!harvey.qualcomm.com!greg
From: [e-mail] (Greg Noel)
Subject: Re: Proposal for default scope
Message-ID: <[e-mail]>
Sender: [e-mail]
Nntp-Posting-Host: harvey.qualcomm.com
Organization: Qualcomm, Inc., San Diego, CA
References: <[e-mail]>
Date: Fri, 8 Jan 1993 22:58:29 GMT
Lines: 27

Donald W. Rouse II <[e-mail]> writes:
>You can accomplish the same thing with
>    class Class {
>    private: static int helper ();
>    public:  int func1();
>    };
>although you don't get the syntactic sugar.

You miss my point.  I specifically want to be able to use a type-safe
helper function that _isn't_mentioned_in_the_class_header_.  That is, the
function is _not_ part of the interface, it's part of the implementation.  A
helper function in the class header is visible, even if it is not accessible,
and can cause an ambiguity under derivation.

Another way of saying this is that there's no reason to declare names that
are part of the implementation in the class header.  If the implementation
is changed, helper functions may be added, changed, or dropped, but as long
as the _interface_ is unchanged, clients of the interface shouldn't need to
be changed.

It also permits multiple implementations of the same interface, which you
could select among based on whatever criteria you wished: speed of operation,
space efficiency, precision of result, freedom from bugs, NIH, ....  If the
implementation functions are fixed as part of the interface, this is not
always possible.
--
-- Greg Noel, Unix Guru         [e-mail]  or  [e-mail]