From owner-freebsd-hackers  Fri Apr 12 01:16:03 1996
Return-Path: owner-hackers
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id BAA10111
          for hackers-outgoing; Fri, 12 Apr 1996 01:16:03 -0700 (PDT)
Received: from rf900.physics.usyd.edu.au (rf900.physics.usyd.edu.au [129.78.129.109])
          by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id BAA10083
          for <hackers@freebsd.org>; Fri, 12 Apr 1996 01:15:53 -0700 (PDT)
Received: (from dawes@localhost) by rf900.physics.usyd.edu.au (8.6.11/8.6.9) id SAA15732; Fri, 12 Apr 1996 18:15:21 +1000
From: David Dawes <dawes@rf900.physics.usyd.edu.au>
Message-Id: <199604120815.SAA15732@rf900.physics.usyd.edu.au>
Subject: Re: DVORAK keyboard drivers
To: jkh@time.cdrom.com (Jordan K. Hubbard)
Date: Fri, 12 Apr 1996 18:15:20 +1000 (EST)
Cc: hackers@freebsd.org
In-Reply-To: <7382.829240624@time.cdrom.com> from "Jordan K. Hubbard" at Apr 11, 96 09:37:04 am
X-Mailer: ELM [version 2.4 PL23]
Content-Type: text
Sender: owner-hackers@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

>> As opposed to knowing cryptic options, commands, and whatnot in /etc/*
>> in order to run UNIX?
>
>Well, two wrongs don't make a right.  There are many who say that the
>stuff in /etc should have a nice front-end tool to encapsulate edits,
>and as much as many people hate AIX I've heard many good things about
>SMIT.  Hackers hate it, of course, but Joe User seems to like it quite
>a bit.
>
>> Get off your high horse. xf86config that does exactly that.
>
>Sorry Kaleb, but xf86config is a creeping undead abortion from the
>foulest pit of Hell.  I would hardly want to hold it up as a shining
>paragon of anything, except perhaps how to construct utterly unusable
>interfaces using shell programming.  You really should give a better
>example to support your argument!

xf86config is all we have right now, and it is *not* a shell script (you
must be thinking of the aborted (in more ways than one) ConfigXF86, which
was a shell script).  We need something better, but I'm happy to have
xf86config rather than nothing.

As to the keymap inheritance, the pre-XKB XFree86 system attempted to do the
inheritance on a key-by-key basis.  The XKB version is geared towards doing
the inheritance on a full keymap basis -- ie, what it wants is the name
of the keymap + any variations.  I don't know if the old key-by-key method
will fit in very well with the philosophy of XKB.

As far as I know (and I could be wrong here), the two main commercial
X servers in this market (the X Inside and Metro Link products), have
the keymamp specified by name in some sort of config file, and don't do
inhertance from the underlying keymap.  This is exactly what XFree86 +
XKB requires.  The only difference that has really been pointed out so
far is in the configuration utilities for creating and changing this
config file (something we are working on right now).  This is a completely
separate issue from the argument between key-by-key inheritance vs the
specification of a keymap by name.  If the console drivers had some concept
of various standard keyboard maps, and could return the name of the keyboard
type/map to the application, XKB could make use of this to set a default
keymap based on what the user has configured for the console.

The argument about ease of configuration is an argument for not using
XFree86, not an argument for not using XKB (which I wouldn't be surprised
to see X Inside provide at some stage, if they don't already).  Please
don't confuse the two.

David