Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!tut.cis.ohio-state.edu!snorkelwacker!ira.uka.de!smurf!urlichs
From: urlichs@smurf.ira.uka.de
Newsgroups: comp.protocols.appletalk
Subject: Re: Interpoll crashes: followup
Message-ID: <1393@smurf.ira.uka.de>
Date: 15 Jan 90 22:10:34 GMT
References: <370@bnrgate.UUCP>
Reply-To: urlichs@smurf.ira.uka.de (Matthias Urlichs)
Organization: University of Karlsruhe, FRG
Lines: 39

In comp.protocols.appletalk aruigrok@bnr.ca (Adrian C Ruigrok) writes:
< In article <1383@smurf.ira.uka.de> urlichs@smurf.ira.uka.de writes:
< > 
< > Find out which machine sends such responses, shut it down, and fire the
< > programmer who caused this kind of behavior.
< 
< You are quite correct!
< I realize that the Macintosh in question should not ever be thinking it 
< has a socket on node 0 (especially since all the other sockets are 
< advertising a correct node in the same NBP packet) but that really is not 
< the point.  Even if this is not a valid response, should Interpoll crash.  
< I think not.  Since its purpose is to detect problems like this, it is not 
< very useful if it crashes before you can figure out what is going on.
< 
That, unfortunately, is a general problem with lots of Apple programs, and of
course almost every other computer in general -- error checking is unnecessary
because programs are assumed to be correct, and if they aren't, it's not the
fault of the OS if those nice little bombs show up.

You are of course right that diagnosis software should not be so unforgiving
of errors it doesn't have control over. I just didn't want to start that type
of argument --- we had it sometime last year...

< As always, the programmer that wrote this did not expect us to change the 
< network from benieth him.  And it was that that caused the break.  It 
< would be a lot easier if we could all see the future.  God!  If we fired
< everyone whose program screwed up 2 years later who would be left? ;-)
< 
If you're talking about having a node zero in this paragraph, it was labelled
as "reserved" all along.
BTW, I don't even know an obvious way to set a node number to zero.
Whoever would want to do that?

If I had that problem here, I would assume it's either a non-Apple box which
is causing this, or someone went poking (not just peeking) around in the
AppleTalk global variables, which is a really serious no-no.

-- 
Matthias Urlichs