Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!iuvax!pur-ee!ea.ecn.purdue.edu!kindred
From: kindred@ea.ecn.purdue.edu (Donald Kindred)
Newsgroups: comp.sys.att
Subject: Re: 3b2/600 Wollogong woes...
Keywords: WIN TCP/IP 2.1
Message-ID: <11314@ea.ecn.purdue.edu>
Date: 1 May 89 16:34:39 GMT
References: <2703@cadre.dsl.PITTSBURGH.EDU> <781@nsscb.UUCP>
Reply-To: kindred@ea.ecn.purdue.edu.UUCP (Donald Kindred)
Distribution: usa
Organization: Purdue University Engineering Computer Network
Lines: 58

In article <781@nsscb.UUCP> det@nsscb.UUCP (Dennis Erwin Thurlow) writes:
>In article <2703@cadre.dsl.PITTSBURGH.EDU> gray@med.pittsburgh.edu (Gray Watson) writes:
>>We are disgusted with teh Wollogong TCP-IP we have for our 3B2's.  The
>>newest version (2.1 I think) is so buggy it crawls around our machine 
>>room leaving a slimy trail.
>>
>Listen up, cause I'm only gonna' say this once!
>I do network support for AT&T here at the hotline.
>We KNOW TCP/IP 2.1 is buggy! That's why we have about
>two dozen fixes for it in our database. We will continue to
>uucp these fixes to 2.1 customers who ask for them even though
>3.0 is now GA.
...
>
>We're not going to rewrite sendmail for you, but if you call
>we CAN help.
>nsscb!det (just a little pissed off about all the bad press :-( )


I too encountered many bugs in WIN TCP/IP 2.1.  In our case we discovered
a few different bugs with rlogin.  When I called up the hotline, they
gave me a choice of no less than 4 versions of a "fixed" rlogin.  Each
different version fixed a different bug; no one version had all the fixes
contained in the other versions.

We also are encountering bugs in the socket compatibility library.
Namely we get an error about a socket operation attempted on a
non-socket, right after the socket had just been "successfully"
allocated to us.  As for AT&T not rewriting sendmail, fine; but
could they please at least provide me with a working socket library
so I can do it myself?  In our case, the sendmail from WIN 2.1 is hosed,
it doesn't work with multiple recipients on remote machines, the socket
library is also hosed, Berkley's sendmail results in the socket on
non-socket error, and AT&T doesn't want to give me the source to
their sendmail.  Net effect, I'm stuck with a sendmail that fails
to deliver mail on 1 out of 3 remote destinations.

To top it all off, I asked if we could automatically receive fixes.
The answer was no.  First we would have to discover the bug ourself,
then report the bug and wait for a fix.  This doesn't work particularly
well in our case, we have a number of 3B2's with WIN TCP/IP 2.1.
My last call to AT&T, I reported over 10 separate problems.  It's
painful having to discover the bug, and report it, only to find that
AT&T was aware of the bug all along.  There's no way to tell how much
mail we simply "lost" because of sendmail.  I do know that we noticed
some messages weren't reaching all intended recipients.

As for trying to get an upgrade to 3.0, who do I need to contact?
I have already contacted my account executive at AT&T.  That was
three weeks ago; I'm still trying to determine the proper person
to contact.

I originally reported my problems with WIN TCP/IP on Jan. 6, 1989
to AT&T's Hotline.  I've talked to both AT&T NJ, and AT&T Lyle IL.
I'm still waiting for some of the "fixes".
-- 
Don Kindred			| Internet:	kindred@ea.ecn.purdue.edu
ACN, Purdue University		| Phone:	(317)494-0120