[comp.protocols.appletalk] System 7 vs. Telnet -- The Real Story

brian@natinst.com (Brian H. Powell) (06/13/91)

     Several days ago, I posted a question about what the truth was behind
the non-MacTCP version of NCSA Telnet not working under System 7.
     At about the same time I was hearing on the net that only the MacTCP
version of Telnet worked under System 7, I was hearing from one of my
users here that he was having no problems with the non-MacTCP telnet and
System 7.  So, I disregarded what I was hearing on the net.
     A few days later, another couple of users reported that they could not
make their telnet work under System 7.  The difference was the telnet that
worked was in a LocalTalk zone off a FastPath 4, and the two who didn't
work were in a zone off a GatorBox CS.  Both routers were configured
similarly.
     Well, after a posting to the net, email to Cayman, and a phone call
to Cayman, I still hadn't gotten anywhere.  A second email to Cayman finally
got a response.  Actually, the second email made something click for a
Cayman engineer, and he found the problem.
     Seems that in the dynamic IP addressing code of telnet, it does an
NBP lookup for "=:IPADDRESS@Zone" (where Zone is specified in config.tel).
Well, the GatorBox responds with a packet that's got ":IPADDRESS@*".  I.e.,
with an entity object that's null.  According to Cayman, the AppleTalk
specification says this is okay.  [I haven't found this documented yet.]
     Well, under System 7, the AppleTalk code drops NBP response packets
with a NULL object name.  Thus, NCSA telnet doesn't see the reponse and fails.
The FastPath puts something in this field, so it doesn't suffer the problem.
Also, MacTCP doesn't do the same sort of dynamic IP addressing, so it doesn't
run into this problem.
     So, the bug, according to Cayman, is really Apple's.

[Time to get out my rusty, trusty asbestos suit, which hasn't been used in
several years.  -- Flame on, set to low:]
     Cayman says they'll fix this "by the end of the summer".  I'm sorry,
I need the fix yesterday, not in a few months.  Hmmm, that FastPath 5 is
looking pretty good.
     Cayman had a good lead over Shiva in the technical support department.
But if Cayman won't fix this soon, they're going to fall behind, and people
like me who buy GatorBoxes because of their good support won't have a reason
to buy a GatorBox any more.
     Come on, Cayman; get with with the program.

Brian