Subject: Re: talk oddity?
To: Danny Thomas <D.Thomas@vthrc.uq.edu.au>
From: Michael G. Schabert <mikeride@prez.org>
List: port-i386
Date: 06/26/1999 18:20:43
>>Hi i386'ers,
>>
>>I don't currently run NetBSD on an x86 machine, but I have it running on an
>>alpha & have had it on my mac68k. I got a buddy of mine to install it on
>>his k6-2 box, & we just tried to use "talk". First, I logged onto his
>>machine & noticed that there was no entry for talk in inetd.conf, so I
>>added it from my alpha:
>>
>>talk dgram udp wait nobody.tty /usr/bin/talk talk
>>
>>where the spaces were tabs (although any whitespace should be fine). After
>>that, I talked my login on his box from my box, & it went through just
>>fine. But any time we'd try to talk with it orginating on his box, we'd get
>>this:
>>
>>[Target machine does not recognize us]
>I think the problem is that Sun designed talk to use host byte-order
>without indicating in the data-stream what that order is. Makes
>inter-operability kinda difficult, though I suppose there could be a hack
>to determine the desired byte-order. I think using network byte-order is
>one of the improvements in ntalk.

I already knew about Sun's broken talk implementations (I hacked my way
around that on my wife's univ account), & I know the differences between
talk and ntalk & that they're incompatible with each other..& I've got a
few utilities that can speak on either port (Peter N. Lewis's talk on MacOS
and ytalk on BSD). I was trying to use talk-to-talk between two NetBSD
machines. It even gave that error when I tried to talk to a user on
localhost.

Mike
Bikers don't *DO* taglines.