🚀 go-pugleaf

In article <1991Jan22.032421.9098@irscscm.UUCP> mfaurot@bogart.UUCP (Michael Faurot) writes:
>I've got a Balance running Dynix 3.14 and would like to know if there's
>a program or means by which one can get both dial-in and dial-out
>useage of a tty port?  Please, no comments about upgrading to a
>Symmetry!  

Don't bother with the Symmetry upgrade (is this comment about an upgrade
acceptable? :-).  We have an S81 running 3.0.17 and Sequent still doesn't
provide for this.  At least if they do, they refuse to tell us about it.

We've tried to get this BUG fixed for some time and have had no success
with Sequent.  I guess they would rather sell more ports and have us buy
more modems.

-- 
-----------------------------------------------------------------------------
Keith Pyle                                UUCP: ...!cs.utexas.edu!execu!keith
Execucom Systems Corp., Austin, Texas     Internet: keith@execu.com
"It's 10 o'clock.  Do you know where      Disclaimer: Everything I say is
   your child processes are?"               true unless I use the word 'the'.
-----------------------------------------------------------------------------

In <31108@sequoia.execu.com> keith@sequoia.execu.com (Keith Pyle) writes:
>In article <1991Jan22.032421.9098@irscscm.UUCP> mfaurot@bogart.UUCP (Michael Faurot) writes:
>>[wants to use ports for dialin and dialout simultaneously]
>>[asks about Symmetry upgrade]
>
>Don't bother with the Symmetry upgrade (is this comment about an upgrade
>acceptable? :-).  We have an S81 running 3.0.17 and Sequent still doesn't
>provide for this.  At least if they do, they refuse to tell us about it.
>
>We've tried to get this BUG fixed for some time and have had no success
>with Sequent.  I guess they would rather sell more ports and have us buy
>more modems.

Well, we certainly like to see people buy more modems :-)

This is not a BUG. It's an opinion about The Way Things Should Be.
Dialin Ports are for dialing in, and Dialout Ports are for dialing out.
No one promised you you'd be able to change configuration dynamicly.

Several people have attempted to be clever and provide programs
that kill off gettys, make lockfiles, edit /etc/ttys on the fly
and kill init. And Sun (and most likely others) have corresponding
dialout devices (/dev/cua#) that do all this in the kernel.

However, even if you do all this, you still have the problem
of modem switch settings. You might want dialin set differently
from the dialout ones.

I personally subscribe to the static philosophy.
We solve our problems by having most people call us.
More hardware often means less software problems.
-- 

	Root Boy Jim Cottrell <rbj@uunet.uu.net>
	Close the gap of the dark year in between

In article <1991Jan22.032421.9098@irscscm.UUCP> mfaurot@bogart.UUCP (Michael Faurot) writes:
> I've got a Balance running Dynix 3.14 and would like to know if there's
> a program or means by which one can get both dial-in and dial-out
> useage of a tty port?  Please, no comments about upgrading to a
> Symmetry!  

In article <31108@sequoia.execu.com>, keith@sequoia.execu.com (Keith Pyle) writes:
> Don't bother with the Symmetry upgrade (is this comment about an upgrade
> acceptable? :-).  We have an S81 running 3.0.17 and Sequent still doesn't
> provide for this.  At least if they do, they refuse to tell us about it.
> 
> We've tried to get this BUG fixed for some time and have had no success
> with Sequent.  I guess they would rather sell more ports and have us buy
> more modems.

One of my customers has purchased a Gpx 6070 from Basic Four which will
be arriving in a few weeks, so I anticipate having to deal with the 
dialin/dialout problem sooner or later.  

Do the fopen() and fcntl() calls under dynix permit you to set and
clear the O_NDLAY flag?

If so, you could use one of the PD versions of uugetty to handle 
incoming calls without preventing folks from using the same 
line as a dialout port.  The uugetty would open the port with 
O_NDLAY clear, thus causing the open() to hang until CD 
goes high on the modem.  When that happens, the uugetty checks 
for the existence of a lock file in usr/spool/locks (or whereever 
locks are maintained under dynix).  If it finds a lockfile, it 
checks the process identified by the PID number in the lockfile 
to see if the process that created the lockfile is still active.  
If not, it "knows" that someone else is using the line for 
dialout.  In this event, the uugetty goes to sleep for periodic 
intervals until the process is no longer active.  If there isn't 
a lockfile, or the PID in the lockfile points to a dead process, 
uugetty behaves like a normal 'getty' program... setting the line
discipline, fetching the user's name, and subsequently exec-ing
'login'.

The potential problem (???) is that open() under dynix may not
hang under certain circumstances.  If so, you could still work
around the problem, but the solution might be messy.

-- 
Russell Lawrence, WP Group, New Orleans (504) 443-5000
russ@wpg.com   uunet!wpg!russ

In article <119792@uunet.UU.NET> rbj@uunet.UU.NET (Root Boy Jim) writes:
>In <31108@sequoia.execu.com> keith@sequoia.execu.com (Keith Pyle) writes:
>>In article <1991Jan22.032421.9098@irscscm.UUCP> mfaurot@bogart.UUCP (Michael Faurot) writes:
>>>[wants to use ports for dialin and dialout simultaneously]
>>>[asks about Symmetry upgrade]
>>
>>We've tried to get this BUG fixed for some time and have had no success
>>with Sequent.  I guess they would rather sell more ports and have us buy
>>more modems.
>
>This is not a BUG. It's an opinion about The Way Things Should Be.
>Dialin Ports are for dialing in, and Dialout Ports are for dialing out.
>No one promised you you'd be able to change configuration dynamicly.

I say tomayto, you say tomaato... (OK, the lyrics don't translate well to
this medium).  One person's bug is another person's feature.

>Several people have attempted to be clever and provide programs
>that kill off gettys, make lockfiles, edit /etc/ttys on the fly
>and kill init. And Sun (and most likely others) have corresponding
>dialout devices (/dev/cua#) that do all this in the kernel.

I'll revise my comment: In my opinion, it is a design deficiency.
Clearly, as you point out, it is possible to have dialin and dialout on
the same port.  Sequent has either chosen not to provide this
capability, perhaps with the thought that they would sell more ports,
or they simply don't care to do anything about it.  Given some of our
experiences with Sequent, I wouldn't be surprised either way.
Then, perhaps they think this is a Good Thing.  Obviously, I disagree.

[Side comment: we've had an S81 for two years and are quite pleased
with the hardware (except for the 8mm tape drive).  However, the two
of us who deal with Sequent on technical matters have fairly large
mail folders and a few hours worth of comments regarding the software
side of the system.  Yes, we've talked with Sequent, up and down the
corporate ladder.  Yes, there have been some improvements, but there
is still a long way to go and I don't feel that they are particularly
responsive.  Yes, I do have some basis for comparison.  Our computer
room presently houses hardware from Sequent, Sun, DEC, HP, Unisys, and
Prime.  We also manage systems running AIX, SCO Unix, and Interactive
386/ix.]

>I personally subscribe to the static philosophy.
>We solve our problems by having most people call us.

If you have the option of deciding that people will generally just call
you, I'm certain that this design is quite acceptable.  However, when
your user base requires both dialin and dialout service, it is not
acceptable.  We have field people who call in to prepare and run demos
and we have local folks who need to call other sites for porting projects.
So, we don't have the luxury that you enjoy and are instead forced into
the position of getting more hardware rather than making more efficient
use of that which we have.

>More hardware often means less software problems.

Ah, but doesn't that lead to more hardware problems? :-)
-- 
-----------------------------------------------------------------------------
Keith Pyle                                UUCP: ...!cs.utexas.edu!execu!keith
Execucom Systems Corp., Austin, Texas     Internet: keith@execu.com
"It's 10 o'clock.  Do you know where      Disclaimer: Everything I say is
   your child processes are?"               true unless I use the word 'the'.
-----------------------------------------------------------------------------

In article <119792@uunet.UU.NET> rbj@uunet.UU.NET (Root Boy Jim) writes:
>In <31108@sequoia.execu.com> keith@sequoia.execu.com (Keith Pyle) writes:
>>In article <1991Jan22.032421.9098@irscscm.UUCP> mfaurot@bogart.UUCP (Michael Faurot) writes:
[stuff deleted]
>Several people have attempted to be clever and provide programs
>that kill off gettys, make lockfiles, edit /etc/ttys on the fly
>and kill init. And Sun (and most likely others) have corresponding
>dialout devices (/dev/cua#) that do all this in the kernel.
>
>However, even if you do all this, you still have the problem
>of modem switch settings. You might want dialin set differently
>from the dialout ones.

I'm in the process of "being clever" at the moment - I'm writing a program
to install in place of uucico.  According to whether the -r1 flag is present,
it will attempt the rigmarole described above, including reconfiguration of
the Trailblazer.  I don't intend to produce a generally-applicable tool, just
a solution to my own immediate problem.

If there is interest in this, I will post details of how well it works,
when it eventually gets finished.

>More hardware often means less software problems.

The trick is persuading people to spend the money :-(

 
-- 
Iain Baird                                        iain@rtf.bt.co.uk
                                        ...uunet!ukc!axion!rtf!iain