From mboxrd@z Thu Jan  1 00:00:00 1970
Date: Fri, 14 Jul 2000 14:25:11 -0700
Message-Id: <200007142125.OAA07055@ohio.river.org>
From: Richard Uhtenwoldt <ru@ohio.river.org>
To: 9fans@cse.psu.edu
Subject: [9fans] command history
In-Reply-To: <200007142103.OAA06763@ohio.river.org>
References: <200007141416.JAA16884@sch1.NCTR.FDA.GOV>
	<200007142103.OAA06763@ohio.river.org>
Topicbox-Message-UUID: ddec1e50-eac8-11e9-9e20-41e7f4b1d025

I just wrote:

>                       instead of rio invoking "rc", rio could invoke
>"ile;rc" where the only effect of the ile command is to bind /dev/cons
>to a fileserver which stores the commandline history and interprets the
>history-editing commands.  that way, no modification to rc or rio would
>be needed: ile would "sit between" rio and rc.

ignore this.  when I wrote this I did not consider that rio sends whole
lines to rc, not individual keystrokes.