Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucsd!sdd.hp.com!decwrl!deccrl!crltrx!zorch.crl.dec.com!jg
From: jg@zorch.crl.dec.com (Jim Gettys)
Newsgroups: comp.windows.x
Subject: Re: keycodes
Keywords: uniqueness of
Message-ID: <5514@crltrx.crl.dec.com>
Date: 7 Jun 90 16:03:50 GMT
References: <36306@sequent.UUCP>
Sender: news@crltrx.crl.dec.com
Reply-To: jg@zorch.crl.dec.com (Jim Gettys)
Distribution: usa
Organization: DEC Cambridge Research Lab
Lines: 36

Each key on the keyboard "should" have a unique keycode, but there is no
guarantee.  For example, on the DEC LK201 keyboard (designed long before
X), you can't tell left shift from right; this is inherent in the
hardware, and cannot be worked around.  The LK401 which we recently
started using on the VT1000 can.  But you can't assume you can
distinguish all keys, and there is no advantage for keycodes to be
shared between keys.  Keyboards should also always report up
transitions properly, and my experience is that implementation of autorepeat
is best left to driver software.  Many keyboards attempt to be too "smart" for
their own good, and end up getting in the way.  So while X is designed
presuming reasonable keyboards, it was not possible to require all
keyboards conform at the time we were designing X.

Another common problem is that some people's keyboards can't report up
transitions on non-modifier keys at all times (or don't due to software
limitations, like the current Ultrix drivers which can't do both
autorepeat and up/down simultaneously; I've fixed this in the
multi-screen driver which will start appearing later this year on our
RISC machines).  

There may also be limitations on what keys can be modifiers (true again
in current Ultrix software, and again fixed in my new driver, which
works as well as the hardware can possibly be used, by telling the
keyboard to do as little as
possible for my "benefit"). 

In general, X asks for "correct" keyboards, but many currently on the
market are inadaquate either due to hardware problems or driver
problems, and so software implementers had better be more careful on
keyboard usage than anywhere else in X.
I use my own company's products only as an example; I have reason to
believe that we are typical of what is on the market today (and we are
fixing as many problems
as possible).

				- Jim