Path: utzoo!attcan!uunet!mcsun!ukc!tcdcs!swift.cs.tcd.ie!maths.tcd.ie!tim
From: tim@maths.tcd.ie (Timothy Murphy)
Newsgroups: comp.std.c
Subject: Re: memcpy
Keywords: memcpy
Message-ID: <1990Sep22.101730.25261@maths.tcd.ie>
Date: 22 Sep 90 10:17:30 GMT
References: <1990Sep19.021418.11574@maths.tcd.ie> <18083@haddock.ima.isc.com>
Organization: Dept. of Maths, Trinity College, Dublin, Ireland.
Lines: 30

In <18083@haddock.ima.isc.com> karl@haddock.ima.isc.com (Karl Heuer) writes:

>In article <1990Sep19.021418.11574@maths.tcd.ie> tim@maths.tcd.ie (Timothy Murphy) writes:
>>In [unix] 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.

>This is incorrect.  The proper way to achieve that effect is
>	memset(buf, c, 21);

I'm afraid I oversimplified my example.
The intention in unix unzip.c was to repeat *any* seqence,
not just a single character, eg
	buf[0] = a;
	buf[1] = b;
	memcpy(buf+2, buf, 20);
is intended to set
	buf[0] = buf[2] = buf[4] = ... = a;
	buf[1] = buf[3] = buf[5] = ... = b;
Incidentally, it's been pointed out to me
that this problem has been fixed in more recent versions
of unix unzip.c .

-- 

Timothy Murphy  

e-mail: tim@maths.tcd.ie