Re: fyi Wollongong TCP/IP for VAX/VMS - Reply to Mark Crispin

Tom Engleman (hpda!hpisoa2!hpindda!unit@ucbvax.Berkeley.EDU)
4 Aug 87 16:34:26 GMT

/ hpindda:comp.protocols.tcp-ip / JERRY@STAR.STANFORD.EDU / 2:29 pm Jul 15, 1987 /
This is Dave Crocker, not Jerry Scott. I recently joined The
Wollongong Group as Vice President, Software Engineering. We will
soon be a host on MilNet, so I have not established an interim
mailbox elsewhere. Please direct any short-term mail to me via
Jerry at this address.

The recent flurry of messages about Wollongong requires a formal
response. As you are aware, The Wollongong Group has been
selling TCP/IP-based products for some years. While we have been
successful in doing so, we have been less successful in maintaining
an unblemished reputation within the Internet community. Recently,
we began taking actions to improve user perceptions. From a technical
standpoint, the most significant of these actions involves upgrades
to our VAX/VMS product called WIN/TCP, especially converting to the
use of 4.3BSD as a code base for the TCP/IP implementation. By doing
so, many long-standing problems were solved and performance has been
substantially improved.

On reviewing the messages that were sent to this distribution list,
it appears that the basis for two of the three explicitly critical
notes was a) system administration errors, and b) the use of very
old software.

At the present time, the new release (3.0) does not have any major
TCP/IP bugs known to us, nor does it crash the operating system. The
immediately previous version (2.3) has not had any bugs that crash
VAXes for a time longer that any Wollongong personnel can remember.

It is our policy to work closely with all users of our products to
satisfy their needs. Mark Crispin's July 6 email message, while it
contained no specific details, has been partially addressed in a
public reply citing cockpit error, rather than faulty software. The
message was sent by a system administrator whose contact with Mark
triggered Mark's note. The system administrator cockpit error we
identified does not involve any software bugs, but it does result in
setting the hosts's own name to a constant ("Unknown"). To eliminate
this confusion, we are changing the software to simply use the text
version of the IP address, whenever a similar administrator error is

As part of a test against one of the systems running Mark's TCP, we
did encounter a client SMTP bug. WIN/TCP 3.1, which will be
released shortly, fixes it. It was only discovered because of high
delay in the Arpanet, thereby causing an extraordinary timeout.

In addition to providing technically competent software, Wollongong must
provide support for our products. This is critical. Although admittedly
flawed in the past, this, too, is being significantly improved, as the
recent TCP/IP activity cited above demonstrates. "Support" is a
separate product and has to be purchased. There have been some customers
who purchased the TCP product but did not, for whatever reason, purchase
support. They then passed on the product to the real end-users and
claimed, falsely, that we would not provide support. The cited case of
our software crashing a VAX cluster appears to be an example of this.
Although we subsequently established direct contact with a portion of
the actual end-users affected in this way, we were unfortunately unable
to find the remainder.

The suggestion about our connecting the the Internet is extremely well-
taken. Part of the reason I was asked to join Wollongong was to bring
some Internet experience in-house. The wheels were already in motion,
I discovered, to get a connection when I came on-board. We were
supposed to be on MilNet about 4 months ago, and are in the final
stages of debugging the telecom link.

Lastly, with regard to our AT&T version of TCP/ should be noted
that we developed this product at the specification of AT&T and we are
not free to add features on our own (AT&T markets the product; we
do not). Hence, please ask them to suggest to us any changes that you
deem appropriate.


This archive was generated by hypermail 2.0b3 on Thu Mar 09 2000 - 14:38:49 GMT