> From: Vojtech Pavlik <[EMAIL PROTECTED]>
> Date: Mon, 1 May 2000 23:41:25 +0200
> To: Jim Gettys <[EMAIL PROTECTED]>
> Cc: Paul Ashton <[EMAIL PROTECTED]>, [EMAIL PROTECTED], [EMAIL PROTECTED]
> Subject: Re: [linux-usb] FW: Hot-plugging USB mice?
> -----
> On Mon, May 01, 2000 at 01:18:30PM -0700, Jim Gettys wrote:
>
> > Arguably, the following is a kludge that should not exist.
>
> The kludge currently exists (not only for X, but also for svgalib and
> GPM and other applications that may want to access a mouse directly),
> and is available on /dev/input/mice, char 13:63. It's a PS/2 style mouse
> device that's always present and sums input from all USB mice.
>
> It should be noted that this indeed *is* a kludge and that I hope that
> it'll be possible to throw it away in the future.
>
> > "Is there some device we can select on that will return if the configuration
> > of the USB bus changes?"
>
> As far as I know there is none. However, I believe that X shouldn't care
> about the USB bus here, but about the input devices themselves. If there
> is anyone I should talk to when designing the Input device interface
> from the X team, I'd love to.
>
>
Consider yourself talking to such people: I've cc'ed Keith Packard on this
mail (who also works for SuSE btw). While I'm not directly concerned (yet)
with USB, I have a project which has the same problem, so Keith and I have
been conspiring on how to solve the problem properly.
> > "Is there some device we can select on that will return if the configuration
> > of the USB bus changes?"
>
> As far as I know there is none. However, I believe that X shouldn't care
> about the USB bus here, but about the input devices themselves. If there
> is anyone I should talk to when designing the Input device interface
> from the X team, I'd love to.
>
>
Such a device is what is needed to make the autoconfiguration of the X
server appropriately dynamic: while we can detect if a device goes away
by ENODEV, we need a way to know when devices appear that the X server
should configure into itself. Polling isn't adaquate for the X server:
we need some way to detect this cheaply.
In fact, arguably, such a mechanism should be independent of bus,
and should let applications know whenever a device comes or goes.
Note that some of these devices aren't strictly input devices: some
of these kinds of input devices also have various LED's and small displays
attached. How important this is, is less clear, but there certainly are
CAD markets which cared enough 8-10 years ago to yell about them. Note
that the X input extension has been around for quite a while (the dates
on it are from around 1992).
--
Jim Gettys
Technology and Corporate Development
Compaq Computer Corporation
[EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
