From owner-freebsd-current  Tue Oct 31 21:22:21 1995
Return-Path: owner-current
Received: (from root@localhost)
          by freefall.freebsd.org (8.6.12/8.6.6) id VAA21442
          for current-outgoing; Tue, 31 Oct 1995 21:22:21 -0800
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19])
          by freefall.freebsd.org (8.6.12/8.6.6) with ESMTP id VAA21422
          for <current@freebsd.org>; Tue, 31 Oct 1995 21:22:11 -0800
Received: (from bde@localhost) by godzilla.zeta.org.au (8.6.9/8.6.9) id QAA16021; Wed, 1 Nov 1995 16:18:58 +1100
Date: Wed, 1 Nov 1995 16:18:58 +1100
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199511010518.QAA16021@godzilla.zeta.org.au>
To: current@freebsd.org, fredriks@mcs.com
Subject: Re: xdrrec_getpos and xdrrec_setpos
Sender: owner-current@freebsd.org
Precedence: bulk

>Subject: xdrrec_getpos and xdrrec_setpos

>	What is the side-effect by having these routines return
>off_t? They both mess with file offsets and call lseek, but currently
>they assume off_t is a long.

They also assume that sizeof(int) == sizeof_long().

I dont know about side effects.  There may be some sign extension bugs
from switching to a signed type.

The mem and stdio getpos and setpos functions are also broken.  The
mem functions assume that size_t is u_int and ths stdio functions
assume that fpos_t is u_int.  The stdio functions might be impossible
to implement if fpos_t is a hairy struct.

Bruce