Xref: utzoo comp.dcom.modems:7145 comp.unix.sysv386:1778 comp.protocols.tcp-ip.ibmpc:4104
Path: utzoo!attcan!uunet!jarthur!elroy.jpl.nasa.gov!ncar!ico!dougm
From: dougm@ico.isc.com (Doug McCallum)
Newsgroups: comp.dcom.modems,comp.unix.sysv386,comp.protocols.tcp-ip.ibmpc
Subject: Re: slip
Message-ID: <1990Nov2.143657.6499@ico.isc.com>
Date: 2 Nov 90 14:36:57 GMT
References: <1990Oct31.115338.4582@nstar.uucp> <1990Nov01.025031.12861@virtech.uucp> <1990Nov01.145738.16101@nstar.uucp> <1990Nov02.032318.18632@virtech.uucp>
Reply-To: dougm@ico.ISC.COM (Doug McCallum)
Organization: Interactive Systems Corp., Boulder CO
Lines: 24

In article <1990Nov02.032318.18632@virtech.uucp> cpcahil@virtech.UUCP (Conor P. Cahill) writes:
...
>If I remember correctly,  SLIP itself does not have error correction.  However,
>the layers above it (TCP in most cases) will usually have thier own 
>error correction.  The main exception to this is the UDP protocol (which is
>used by NFS).

A bigger problem than not having error correction is that SLIP has no
error "detection".  The checksumming used by IP and UDP is very simple.
Its a one's complement checksum that cannot detect certain classes of
errors.  For example, suppose one byte in the data stream has a 1 bit
changed to a 0.  Suppose an even number of bytes later a different
byte has a 0 bit changed to a 1 in the same position as the previous
error.  If nothing else has changed, IP or UDP (or TCP) won't see an error
and you have corrupted data.  This usually doesn't hurt for a telnet or
rlogin session, but it isn't particularly nice for NFS or file transfer.
Undetected data corruption is not a desirable trait but it is something
that occurs with SLIP.

PPP has a CRC mechanism so will detect errors.

Doug McCallum
Interactive Systems Corp.
dougm@ico.isc.com