Path: utzoo!attcan!uunet!mcsun!unido!tub!opal!fub!geminix!gemini
From: gemini@geminix.in-berlin.de (Uwe Doering)
Newsgroups: comp.dcom.modems
Subject: Re: USR 9600 baud dual standard
Message-ID: <54AKM7C@geminix.in-berlin.de>
Date: 12 Oct 90 11:27:28 GMT
References: <1990Oct8.195418.1817@cbnews.att.com> <2461@van-bc.wimsey.bc.ca> <2170@jwt.UUCP> <2509@van-bc.wimsey.bc.ca>
Organization: Private UNIX Site
Lines: 43

sl@van-bc.wimsey.bc.ca (Stuart Lynne) writes:
>In article <2170@jwt.UUCP> john@jwt.UUCP (John Temples) writes:
>>In article <2461@van-bc.wimsey.bc.ca> sl@van-bc.wimsey.bc.ca (Stuart Lynne) writes:
>>
>>>van-bc run's 2 PEP modems with baud rates locked at 19.2. With no flow
>>>control.
>>
>>How can you pump data at 19.2k bps into a modem which can only
>>transmit it at 17-18k bps without using flow control?
>
>With uucp you only ever send a maximum of 220 bytes of data before stopping
>and waiting for an ack packet. The modem will not send an ack when it no
>longer has enough buffer space. But it always can receive without error the
>three packets (220 bytes) that you will send without acknowledgement.

Stuart, I think you didn't read the original article of this thread very
well. The statement about uucico needing hardware flow control refered to
the `e' protocol. This has nothing to do with the `g' proto. `e' proto is
a streaming protocol and doesn't have the windowing technique that `g' proto
is using. If only works over error free channels that do their own
flow control. Therefor your computer as well a the modem has to have
hardware flow control.

`e' proto was developed for LANs. Due to this design goal most
implementations in use today have a flaw that makes it unreliable
over serial line. It simply gets stuck under certain conditions.
This has to do with the message receiving function assuming that
messages will arrive in one single packet (which is true for LANs),
whereas error free modems do their own packetizing which usually
confuses the uucico. It took months until I discovered the reason
for this instability. I don't know whether this is fixed in newer
releases of HDB UUCP. At least in ISC 2.0.2 it isn't fixed. Don't
know about ISC 2.2.

Therefor, try it out if you have `e' proto but don't be disappointed
if it doesn't work. Note that both sides need a clean `e' proto.
This makes it even less likely that you can use it in the real world. :-(

      Uwe
-- 
Uwe Doering  |  Domain   : gemini@geminix.in-berlin.de
Berlin       |----------------------------------------------------------------
Germany      |  Bangpath : ...!unido!fub!geminix!gemini