Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ukma!psuvm.bitnet!psuecl!wcf
From: wcf@psuecl.bitnet (Bill Fenner)
Newsgroups: comp.mail.uucp
Subject: Re: The login: prompt.
Message-ID: <22266@psuecl.bitnet>
Date: 13 Mar 89 00:20:49 GMT
References: <0724.AA0724@julie>
Followup-To: comp.mail.uucp
Organization: Engineering Computer Lab, Pennsylvania State University
Lines: 56


In article <0724.AA0724@julie>, mcr@julie.UUCP (Michael Richardson) writes:
>   I am in the process of writing a "Front Door/Sea Dog/Binkley"-like
> program (for the Amiga of course) that sits in front of a bbs and does mail
> transfer sessions with non-human callers. In Unix terms we are talking
> about a getty(8) with built in X and Z modem.
>   I suspect that it would simply make sense to put up---
> "UUCP?" and if I receive a "Y" then print a "login:" otherwise
> drop into the bbs (after checking for the fidonet [yuck---Zmodem excepted]
> protocols)
>
>   Does anyone know if this is going to present any big problems to any
> known sites that I might care about? (i.e. IBM 360 running a uucp written
> in JCL... :-))
>
>
>   My "vision" is of a local dialup network of computers that would be connected
> together, making everyone a "point" (none of that dial/budy/autodial stuff)
> and also to have connections to the outside world. (The problems of what
> this would do to net bandwidth has not been overlooked. I can imagine
> subscribing to a group and then subscribing to individual threads --- like
> a glorified kill file. Also, Distribution: would be a hard and fast
> rule!!!)
>
>   I've done enough uucp connection configuring (but never with HDB uucp,
> only BSD or mickey mouse micro uucp stuff.) to realise that most
> system administrators already know how to do "ogin:" <name> "sword:" <password>"
> stuff, but is there any real reason that this has to be done?
>
>   Under Fidonet, a (horrible kludge) _special_ character (0xAE I think) is
> sent to start the base level protocol. (No security) When someone came
> up with the idea of doing protocol negotiation, a new _special_
> sync character was defined (of course that is extendible---but only with
> a sixteen bit mask.)
>
>   I don't want to put up a "login:" prompt and then look for sync's
> or login names, confusing the heck out of human callers when they have to
> enter "bbs" or they enter their name and then I pull the bbs (or whatever---
> I'm not specific) up. I don't to have anything to do with "user logs" or
> passwd files.

Why not do it the way Binkley does it -- wait for an escape or some other
key from the user to load the BBS, wait for the Fido sync characters, or
wait for some other string, specified by the user.  Mine is UUCPPLEASE.
 My L.sys entry is
hogbbs Any ACU 2400 8142389633 econds--econds UUCPPLEASE UUCP-UUCPPLEASE-UUCP
The bbs says something like
Press your escape key now, or wait 25 seconds
that's where the econds comes from.  Then when it gets a UUCPPLEASE, it
says
Loading UUCP, please wait... thus the UUCP-UUCPPLEASE-UUCP.

It's worked well for several months.

  Good luck!

  Bill Fenner