[Simh] Serial Console

[Johnny Billquist]

On 2012-07-03 13:38, Mark Benson wrote:
> On 3 Jul 2012, at 12:03, Johnny Billquist <bqt at softjar.se> wrote:
>
>> On 2012-07-03 11:49, Mark Benson wrote:
>>> Simple question: Can SimH's CONSOLE output be directed to a Serial port
>>> (i.e. via /dev/ttyS0)?
>>>
>>> If so how does it handle waiting for a terminal connection?
>>>
>>> If not, is there a reason why?
>>
>> I think I looked into this a few years ago, and found that it can't. There is no absolute reason why it would not be able to, but exactly how you control a physical serial port differs between OSes, so I suspect it would be hard to write generic.
>
> Ah, that makes sense, yes, different OSs use different device names
> etc. even within common UNIX methods. This wouldn't be such a big
> hurdle if the onus was on the user to provide the full device path
> though. That, of course, doesn't help with Windows or VMS but it might
> widen it's UNIX campatibility?

It's not just device names. That part would actually be the easy thing. 
You just give the path in the config file. No different than how you 
point out disk images for disks.

However, how do you change the speed of a serial port? Or detect the 
modem signals? Or any other direct manipulation? That is way different 
on different systems.

You asked initially about the console, which is often a little special 
case, since that port in many systems are a little more primitive, but 
in general the same problems apply.

>> However, I think that under Unix, if you just log in on a serial port and start simh, you'd be more or less there already?
>
> No, you see what I would like to do is connect a VT terminal directly
> to SimH to use either as the OP console or to provide a real VT access
> for the DZV11 emulation. This is all about cutting out multiple layers
> if terminal emulation to get real VT interaction direct to the
> emulator.

Well, my answer was about the console specifically. If you are looking 
for multiply serial ports being emulated in simh, and mapped to real 
physical serial ports on the host, then you need a "proper" solution.

> The reason is I have noticed that over Telnet VMS is pretty robust but
> only offers a 'cut down' cinnection to remain compatible. Conversely,
> GNU screen working directly as the console convinces VMS that it's a
> real VT then it pukes when you run something like EVE or
> TCPIP$CONFIG.COM because VMS is assuming the terminal can handle
> features that screen cannot. There are also other things like
> softloaded fonts etc that only real VT terminals handle properly that
> break ir don't work in screen or telnet.

Well, get a proper terminal emulator... :-)
Apart from the soft font thing, I'd say that xterm is what you want. 
It's the only terminal emulator that I have found which actually emulate 
a VT100, VT102 or VT220 correct. But you do not have soft fonts...

telnet by the way, should not be confused with a terminal. It is not. 
telnet is merely the program that you use to connect to a system. You 
can compare it with a cable. A cable is not a terminal. The terminal 
sits at one end of the cable.

screen on the other hand do emulate a terminal. Not very good, though.

> It'd be nice to hook the VT510 up and use that instead... wasjyst a thought.

Grab your VT510, connect it to your Unix system on a serial port. Log in 
to Unix. Telnet into simh, and voila!

But apart from that, sure, I agree that direct connections between simh 
and physical serial ports would be nice.
And like I said, it is not impossible. You just need the code, and the 
code will have to be very different depending on the OS.

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt at softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol