Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site brl-tgr.ARPA
Path: utzoo!linus!philabs!cmcl2!seismo!brl-tgr!gwyn
From: gwyn@brl-tgr.ARPA (Doug Gwyn <gwyn>)
Newsgroups: net.lang.c
Subject: Re: (void *)
Message-ID: <203@brl-tgr.ARPA>
Date: Fri, 26-Jul-85 14:31:50 EDT
Article-I.D.: brl-tgr.203
Posted: Fri Jul 26 14:31:50 1985
Date-Received: Sun, 28-Jul-85 08:04:17 EDT
References: <1035@homxa.UUCP> <921@umcp-cs.UUCP> <2462@sun.uucp> <318@baylor.UUCP>
Organization: Ballistic Research Lab
Lines: 14

> I just realised a major problem with (void *). On machines like the HP-1000
> a pointer to a byte and a pointer to a word don't have the same bit-pattern.

This is true of most word-addressed machines.

> How are you supposed to deal with this without having a terribly in-efficient
> implementation of 'C'.

Why is this a problem?  That is what (void *) is specifically meant to
deal with.  Assigning to or from a (void *) datum on such a machine will
involve mapping to/from a canonical pointer representation, but that is
what (char *) is currently used for anyhow.  Presumably a (void *) would
look just like a (char *).  The main reason for different types is to
allow meaningful stricter type-checking, especially of malloc() values.