Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!seismo!think!rlk
From: rlk@think.COM (Robert Krawitz)
Newsgroups: comp.windows.x
Subject: Re: Gnuemacs & X
Message-ID: <9159@think.UUCP>
Date: Sun, 11-Oct-87 21:20:36 EDT
Article-I.D.: think.9159
Posted: Sun Oct 11 21:20:36 1987
Date-Received: Wed, 14-Oct-87 00:52:58 EDT
Sender: news@think.UUCP
Reply-To: rlk@THINK.COM
Organization: Thinking Machines Corporation, Cambridge, MA
Lines: 25

In article <26900007@uiucdcsm> grunwald@uiucdcsm.cs.uiuc.edu writes:
]
...updates don't always happen on suns
]
]I suspect some threshold for gathering & batching updates to the display
]server. Any clues?

Nope.  It's a kernel bug, in that SIGIO's don't get sent properly on
sockets.  The display code itself uses SIGIO in various nefarious
ways, and these apparently don't get fired properly.

Fixed in all 4.3 type systems I know of.

BTW, does anyone else using X need asynchronous I/O?  GNU Emacs does
to make C-g work (this is an asynchronous interrupt) and the display
code uses it for efficiency (it waits a few seconds for input before
redisplaying, so it has to listen for that much time while sleeping;
this is what I believe causes the redisplay problem).  Ralph Swick and
I started to discuss this issue last year but we never got very far...

cca >>>>>>>>>>  |
harvard >>>>>>  |
bloom-beacon >  |think!rlk	Robert Krawitz <rlk@think.com>
rutgers >>>>>>  |
ihnp4 >>>>>>>>  .