Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/5/84; site kitty.UUCP
Path: utzoo!linus!philabs!cmcl2!seismo!rochester!rocksanne!sunybcs!kitty!peter
From: peter@kitty.UUCP (Peter DaSilva)
Newsgroups: net.micro.pc
Subject: Re: A New Compiler Model
Message-ID: <133@kitty.UUCP>
Date: Thu, 18-Jul-85 11:58:26 EDT
Article-I.D.: kitty.133
Posted: Thu Jul 18 11:58:26 1985
Date-Received: Sat, 20-Jul-85 12:37:24 EDT
References: <368@kontron.UUCP>
Distribution: net
Organization: Recognition Research Corp., Clarence, NY
Lines: 21

> There are several different memory segmentation models on the 8086 family:
> 
>         small           64K code, 64K data
>         medium          many 64K code segments, 64K data segment
>         large           many 64K code segments, many 64K data segments

Actually small model on microsoft & lattice compilers is 64K total code
and data, as you suggest. If you're careful the resulting file can be
exe2bin-ed with no problems. I wish lattice supportted 64K code plus
64K data (split I&D)... it currently has the following:

	small	64K code+data
	prog	LARGEK code + 64K data
	data	64K code + LARGEK data
	dataS	64K code + many 64K data
	large	LARGEK code+data.
	largeS	LARGEK code + many 64K data

LARGEK isn't implemented as a lot of segments, but rather as 32 bit pointers
of some non-contiguous form. It looks like regular 20 bit pointers except
you can't pretend they're longs.