Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!mnetor!uunet!seismo!husc6!cmcl2!rutgers!sri-spam!ames!ucbcad!ucbvax!SIMTEL20.ARPA!WANCHO From: WANCHO@SIMTEL20.ARPA ("Frank J. Wancho") Newsgroups: comp.protocols.tcp-ip Subject: Wollongong TCP/IP for VAX/VMS Message-ID:Date: Tue, 14-Jul-87 04:00:00 EDT Article-I.D.: SIMTEL20.WANCHO.12318234098.BABYL Posted: Tue Jul 14 04:00:00 1987 Date-Received: Wed, 15-Jul-87 04:22:12 EDT Sender: daemon@ucbvax.BERKELEY.EDU Distribution: world Organization: The ARPA Internet Lines: 52 I have a somewhat different viewpoint and solution to the ongoing commentary concerning the Wollongong implementations of TCP/IP and supporting software for various operating systems. First, let me point out that I get a different set of complaints than Mark gets. As the postmaster for this DEC-20 site, which is the origin/relay point for several large mailing lists, I get a certain set of complaints from the postmasters at BITNET sites who are having problems with our headers. What has made the difference is that in most cases, I have been able to deal directly with the authors of the software in question to resolve the problems in interpretation of the RFCs using our real-world (Internet) messages. What is different with dealing with users of Wollongong software is that they are in the position of having to report problems into a corporate environment which has never had to interface their software into a large heterogenous network such as ours. In house, they test their software against other implementations of their own software, and it's kinda hard to duplicate a problem, much less be aware that a problem exists in that situation. Recall the early days surrounding the rapid implementation and heterogeneous testing of various TCP/IP implementations just a few short years ago and you'll understand my point. The solution is obvious: The Wollongong Group should have a host on the Internet so that they can find and fix problems before their customers do, among other things. This is not without precedent. Not too long ago, when the predominant operating system on the net was TOPS20, DEC had, and still has one or more of their own TOPS20 hosts on the net, testing their TCP/IP implementations (as was BBN testing their versions). I'm sure there are other examples, and I would suppose that there were and still are other reasons for DEC to be on the net. In a recent analysis I made of the various operating systems listed in the NIC HOSTS.TXT file, by far the most predominant was Unix, in various flavors on various machines. Those hosts are mostly running the 4.xbsd version, with Berkeley certainly represented directly on the net. The second was VMS systems, presumeably with a majority of them running Wollongong software. Well, it appears such a Wollongong host does exist, according to the NIC HOSTS.TXT file and the WHOIS database: TWG.ARPA, 26.5.0.73. However, it appears to be non-operational or a reserved designation. At least I have not been able to get a response from that host, yet. I firmly believe that the sooner they get on the net as an operational host, we will see a significant and radical improvement in the situation. Anything that can be done to speed up their connection would be of great benefit to all of us. Note carefully: I'm not necessarily advocating that only by virtue of having a TCP/IP software package should every developer have a host on the Internet. Such developers should at least adopt a host for in resident beta tests. I suspect that every major developer already has such a connection, except Wollongong... --Frank