Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!ut-sally!utah-cs!sandra
From: sandra@utah-cs.UUCP (Sandra J Loosemore)
Newsgroups: comp.sys.atari.st
Subject: Re: MicroGnuEmacs
Message-ID: <5048@utah-cs.UUCP>
Date: Fri, 9-Oct-87 18:42:00 EDT
Article-I.D.: utah-cs.5048
Posted: Fri Oct  9 18:42:00 1987
Date-Received: Sun, 11-Oct-87 16:41:48 EDT
References: <8710071414.AA01109@work3.icase> <168@decvax.UUCP> <5619@j.cc.purdue.edu>
Organization: University of Utah CS Dept
Lines: 20
Keywords: mg emacs scrolling

I've already replied to the author of this via e-mail but since there
seems to be general misunderstanding as to what is going on, I will
try to clear it up.

MG's display code is based on Gosling's algorithm, which was really
designed to work with terminals that support scrolling regions, like
the VT100.  The ST's Cconout() function doesn't support scrolling
regions, but (like an H19 and several other terminals), it does have
insert line and delete line functions.  You can simulate scrolling
regions with these functions, but it does give you that "bouncing mode
line" problem.  Making the cost of scrolling very high would probably
make it act even dumber, like a real VT52 where you have to do a
clear-to-end-of-page and redraw *everything* to get the same effect.

The real solution is to fix MG to bypass Cconout () and manipulate
the screen directly.  I've been given some code to do just that and if
I have time, I'll try to get it integrated into the next release.

-Sandra Loosemore
sandra@cs.utah.edu, sandra@utah-cs.uucp