[comp.sys.encore] Annex problems: intr flush, logout message, no connection drop on dcd

rayan@ai.toronto.edu (Rayan Zachariassen) (02/04/89)

I'm trying to beat an Annex into shape to take over as our terminal server,
but am running into some problems... can anyone help? (Annex-UX R4.0)

1.	dialing into a modem whose interface is wired at 19.2kb,
	rlogin'ed to a SunOS 4.0 box, output isn't flushed on interrupt.
	Dialin baud rate 2400, sun pty baud rate originally 19.2kb,
	but changing it to 2400 has no discernible effect.  Reading
	through the info-encore archives it seems this problem was solved
	with the switch to 4.3 networking with R2.1...  in fact, I seem
	to remember testing this in some situation (hardwire line?) and
	it working, but I'm not sure (and can't try now).  Ideas?

2.	running with security enabled, all seems to work nicely except
	for logout messages.  In acp_policy.c, I just added a call to
	outputstring() before the acp_acknowledge() in annex_to_net_logout()
	and port_to_annex_logout().  I never see these strings, although
	the log file entries appear nicely.  What am I doing wrong?

3.	if you dial out through a modem on the annex (via the port server),
	and the remote modem drops carrier causing local modem to drop DCD,
	the network connection to rtelnet doesn't die.  Why?

4.	I had to change rtelnet so that it wouldn't immediately open a
	connection (kind of inhibits using a bidirectional modem for dialin),
	but wait until there was data on the pty.  When rtelnet terminates
	the connection it sends a TELOPT_TM cookie, or some such, and
	apparently negotiates connection shutdown.  The problem is that
	after the connection dies, the modem sends junk to the annex which
	gives it a cli (meaning you can't dial out...).  I had to put a
	call to reset_lines() inside rtelnet, but I would have thought the
	behaviour would be more graceful about these things, no?  Also,
	sometimes rtelnet would hang inside send_tm() after turning off
	NBIO on the net fd.  This situation indicates that the modem went
	away (I see "NO CARRIER") but the connection wasn't dropped (see 3).

Apart from this, the box and software behaved fine once I fixed the
portability bugs in na.

Ideas/pointers would be appreciated.

rayan