[comp.dcom.lans] Problem with running "rn" over Starlan

alexb@cfctech.UUCP (Alex Beylin) (12/09/88)

A few days ago we have installed AT&T Starlan-10 network.
It is suppose to follow all the OSI rules.  Hardware is 
10 MBit twisted-pair Ethernet.

One of the things that we could not get to work over the
net is "rn".  It dies as soon as it tries to go in the 
cbreak mode.  Now, that is fully understandable as rn
expects a "tty" port and starlan-10 is talking to it over
some "slan" device rn knows nothing about.

Before we jump and start modifying rn code and make 
ourselves incompatable with the next revision of rn,
does anyone have a simple answer or a patch for this 
problem?  Does anyone have rn running over Starlan
(old or new).

Thanks in advance for any input.  This is cross-posted
to news.misc and comp.dcom.lan, but I am directing follow-ups 
to comp.dcom.lan as a more appropriate one.

	

============================================================
 Alex Beylin (alexb@cfctech.uucp)
 {umix,mailrus}!cfctech!alexb	
------------------------------------------------------------
 DISCLAMER:  Any opinions expressed here are are my own and
 may be contrary to polices or procedures of my employer.
============================================================

matt@iquery.UUCP (Matt Reedy) (12/14/88)

In article <518@cfctech.UUCP>, alexb@cfctech.UUCP (Alex Beylin) writes:
> 
> A few days ago we have installed AT&T Starlan-10 network.
> 
> One of the things that we could not get to work over the
> net is "rn".  It dies as soon as it tries to go in the 
> cbreak mode.  Now, that is fully understandable as rn

We had a discussion about this a few weeks ago in the 3B2 mailing list.  The
Starlan documentation (STARLAN NETWORK AT&T 3B2 Computer User's Guide) page
3-8 states that the Starlan TTY Server doesn't support raw character proc-
essing (and a couple of other termio(7) features).  The final resolution
stated by someone was that V.3.2 has corrected this problem, but I haven't 
verified this myself.  Has someone found this to be true?

matt
-- 
Matthew Reedy                 UUCP: {harvard!adelie,gatech!petro}!iquery!matt
Programmed Intelligence Corp.
400 N Loop 1604 E, Suite 330
San Antonio, TX  78232        (512) 490 6684

jos@idca.tds.PHILIPS.nl (Jos Vos) (12/28/88)

In article <518@cfctech.UUCP> alexb@cfctech.UUCP (Alex Beylin) writes:

>One of the things that we could not get to work over the
>net is "rn".  It dies as soon as it tries to go in the 
>cbreak mode.  Now, that is fully understandable as rn
>expects a "tty" port and starlan-10 is talking to it over
>some "slan" device rn knows nothing about.

We did have the same problems using rn with a V.3 streams terminal.
Is that the case at your site?

The V.3 read(2) returns -1 when the O_NDELAY flag is set on a streams tty
and there is no data, i.s.o. 0 what is done with "normal" tty's.
I did make 3 very minor changes to the code of rn to let it work properly
on all devices. 
If you are interested, I can find them in my "history".

-- 
-- ######   Jos Vos   ######   Internet   jos@idca.tds.philips.nl   ######
-- ######             ######   UUCP         ...!mcvax!philapd!jos   ######