lkml.org 
[lkml]   [1998]   [Jun]   [26]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: uniform input device packets?
Date
From
Dossy wrote:

> On 1998.06.24, Pavel Machek <pavel@elf.ucw.cz> wrote:
> > Hmm, I do not like this one. What is wrong with attaching 9 mice to my
> > PC?
>
> You currently can't attach 9 mice to a standard PC, so what's the
> problem? At least my proposed methodology would *possibly* support
> 8. With USB, 9 mice on your USB might be *possible*, but highly
> unlikely.

What about a box with a Stallion card and mice plugged into it,
to provide all the mice for a many-headed X server (eg, one with
2 or more PCI busses almost fully populated with both PCI and ISA
video cards) or for an X server farm using the gpm named pipe and
some network sockets to share the mouse over the network?

If you don't think the idea of a 9-headed X server with a Stallion
card providing the mouse ports is possible due to CPU limits, just
think about a 4 x Pentium II 400 on a dual-PCI bus board (plus ISA).
I expect it would work acceptably; I used to run X on my 386 and on
machines of non-Intel architecture with much less CPU power.

Or joysticks - even my Playstation has games which support more than 8 players
(controller sharing + multitaps to do it, though, but the point is it's
possible and people even write software which does it). Someone might have
a big-screen display and want a game with a dozen or so gamepads at a
shopping centre, whatever.

A hard limit on the number of any type of device is not good, especially
if you make input packets have to fit into a machine word which makes
extending the hard limit difficult (look at how long the 32 signal limit
stayed around, causing problems with threading in apps which used SIGUSR1
for a user signal, such as squid, under the kernels with that limit).

USB supports something like 128 devices, there are also SCSI-based input
devices you may want to incorporate into your scheme and so on. You can
never predict the configurations where people might want to use Linux
boxes.

I really don't like the idea of tagged input packets to begin with.
What's the problem with a /dev/keyboard[#], /dev/joystick[#],
/dev/mouse[#] and so on? Or even /dev/input[#] (which I don't particularly
like the idea of). It's far more generic. It's simpler. It doesn't need
a bloated inflexible protocol. Any application simply connects and gets
only the information it wants.

David.


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu

\
 
 \ /
  Last update: 2005-03-22 13:43    [from the cache]
©2003-2011 Jasper Spaans