jbvb@FTP.COM ("James B. Van Bokkelen") (02/28/91)
Archive-name: internet/msdos/harvard-pcip/1991-02-27
Archive-directory: husc6.harvard.edu:/pub/pcip/ [128.103.1.56]
Original-posting-by: jbvb@FTP.COM ("James B. Van Bokkelen")
Original-subject: Re: Recent Dev. of PC/IP
Reposted-by: emv@ox.com (Edward Vielmetti)
Some random points:
PC-IP's multiple-connection TCP was written at MIT long ago, but never really
debugged until Dan Lanciani included it in the most recent Harvard version.
Unless it's changed a lot from what we began with, it is slow and lacks many
important features like adaptive retransmit.
The Packet Driver interface for PC-IP was done by Karl Auerbach while he was
doing consulting work for TRW.
Drew Perkins at CMU was responsible for the port to MSC 3.0. I believe the
Harvard version is built with MSC 5.1. PC-IPs have always been small-model
(the tasker and the ASM glue routines only understand that memory model).
Wollongong shipped various PC-IP derived products, first based on the MIT
version, later on the SU version. Bridge also shipped SU-derived products,
but I don't think current DOS TCP/IPs from either TWG or 3Com have much of
any PC-IP heritage. Sun's PC-NFS had a little PC-IP in it at one point.
IBM's TCP/IP for DOS was an earlier version of the UMD code, closer to its
CMU ancestry. IBM has also followed a line of development based on an
early port of the MIT PC-IP to Xenix through to their current AIX TCP/IP.
James B. VanBokkelen 26 Princess St., Wakefield, MA 01880
FTP Software Inc. voice: (617) 246-0900 fax: (617) 246-0901