Newsgroups: comp.std.c
Path: utzoo!henry
From: henry@zoo.toronto.edu (Henry Spencer)
Subject: Re: memcpy
Message-ID: <1990Sep21.031142.23718@zoo.toronto.edu>
Organization: U of Toronto Zoology
References: <1990Sep19.021418.11574@maths.tcd.ie>
Date: Fri, 21 Sep 90 03:11:42 GMT

In article <1990Sep19.021418.11574@maths.tcd.ie> tim@maths.tcd.ie (Timothy Murphy) writes:
>In unzip.c it is assumed that the effect of
>	buf[0] = c;
>	memcpy(buf+1, buf, 20);
>is to set
>	buf[0] = buf[1] = buf[2] = ... = buf[21] = c.

The effects of invoking memcpy with overlapping operands have never been
officially promised, that I recall.  ANSI C explicitly says that the
implementation is allowed to assume that they do not overlap, and it
is the programmer's responsibility to assure this.

This sounds like rank-amateur code.  If he wants to set all characters
in a buffer to the same, memset() will do that efficiently and cleanly.
Even if an overlapping memcpy() works, it is not going to be particularly
fast.
-- 
TCP/IP: handling tomorrow's loads today| Henry Spencer at U of Toronto Zoology
OSI: handling yesterday's loads someday|  henry@zoo.toronto.edu   utzoo!henry