Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 8/7/84; site ucbvax.ARPA
Path: utzoo!watmath!clyde!burl!ulysses!mhuxl!ihnp4!ucbvax!info-vax
From: info-vax@ucbvax.ARPA
Newsgroups: fa.info-vax
Subject: Re: Networking query
Message-ID: <1992@ucbvax.ARPA>
Date: Mon, 17-Sep-84 08:58:43 EDT
Article-I.D.: ucbvax.1992
Posted: Mon Sep 17 08:58:43 1984
Date-Received: Tue, 25-Sep-84 04:05:20 EDT
Sender: daemon@ucbvax.ARPA
Organization: University of California at Berkeley
Lines: 24

From: Richard Garland 

I have had no problems losing characters.  Trick is to insure that you
have plenty of type-ahead buffer space and use flow control.  In the case
of terminal emulation (passthrough) flow control is usual to have anyway
(since for example VT100's use it in both directions).  When something
like KERMIT is running, flow control is generally turned off and KERMIT
does the packet retries.   I have certainly run these things routinely
at 1200 baud with a dialer, and at 9600 (to an RT11 system rather than
Unix) direct wired.   I would stress unless the outgoing port is set
with type-ahead, and HOSTSYNCH (I think that's the one) there is no
chance.

Another approach I have been told works, is to use Unsolicited IO Mailbox
AST's and read the whole buffer in rather than single character IO. In
that case you just set up a Mailbox AST and wait.   When it comes you can
ascertain the buffer size and you just read it all at once.   Obviously
Type-ahead must be on here as well.

Note the type-ahead buffer size can be increased arbitrarily with the
alternate type-ahead setting.

					Rg
-------