Path: utzoo!utgpu!news-server.csri.toronto.edu!clyde.concordia.ca!uunet!munnari.oz.au!metro!macuni!mqcomp!s8925188
From: s8925188@mqcomp.mqcc.mq.OZ (Philip Craig)
Newsgroups: comp.dcom.modems
Subject: Re: Is MNP5 compression better on all-caps?
Summary: MNP5 is better with smaller character sets
Keywords: MNP5,compression
Message-ID: <402@macuni.mqcc.mq.oz>
Date: 20 Aug 90 04:16:10 GMT
References: <14862@drilex.UUCP>
Sender: news@macuni.mqcc.mq.oz
Reply-To: s8925188@mqcomp.mq.oz (Philip Craig)
Organization: Macquarie University, School of Mathematics, Physics, Computing and Electronics
Lines: 23

In article <14862@drilex.UUCP> dricejb@drilex.UUCP (Craig Jackson drilex1) writes:
>When working with our mainframe, which likes to output in all caps,
>the choppiness seems to be a good deal less.  I remember that there are
>people out there who are familiar witht the MNP5 scheme--is it really
>compressing more on all-caps?  This would be a useful thing to know.

MNP5 maps the most frequently occuring characters into tokens that have
a short number of bits, and less frequent characters into tokens of longer 
length. The shortest token is 4 bits, and the longest is 11 bits.

When a machine talks only in all caps, then the character set it is using
has 26 less characters in it. This means that other characters can move
up the table and use the shorter length tokens.

The compression has nothing to do with the all-caps nature, but the one-case
nature is what is important. The same effect would be achieved if the machine
used only lower case (and it'd be a lot more readable, I reckon!)
-----------------------------------------------------------------
Are you sure you want to remove SUN microsystems [yes/no]?

ACSnet          ( s8925188@mqcomp.mq.oz )
UUCP            ( uunet!munnari!mqcomp.mq.oz!s8925188 )
Internet        ( s8925188%mqcomp.mq.oz@uunet.uu.net )