[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Input team resurrection




On 09-Jun-99 Christian Reiniger wrote:
> Stephane Peter wrote:
> 
>>> But it's surely better to extend libgii & build some app around it than to
>>> rewrite everything from scratch.
>>
>>Why not. You can as well use the existing libGII in that daemon after all if
>>it serves the purpose.
> 
> That's what I meant ;)

By the way, I've read the libGII docs and it looks like it would rather be the
kind of library that would sit on top of such a daemon/API. i.e. it would pump
events from it and present it to the end-user in a generic way through the GII
event queue.

>>> Yup, but if your system should be capable of handling multiheaded setups
>>> you need some controlling code in the kernel.
>>
>>I don't see why we should bother about multihead in mouse/joystick drivers ?
> 
> Well, IMHO multiheaded setups will become more common (the G400 already has
> two monitor ports for example), so if we do want to do it properly we have
> to take that into account.

Yeah, but the main issue that need kernel support for multihead is for the
graphics hardware (managing multiple cards at the same time). Input devices are
a different issue, and probably much easier ;). That's what I've understood
after working at SciTech and watching the XFree86 developer mailing list ;)

>>Even though several of those may be plugged at the same time, I don't see why
>>you would need any form of special kernel support for that.
> 
> Well, some instance has to do the device-to-(virtual)console mapping. And
> the only instance knowing enough and having enough control is the kernel.

That would make sense for several devices associated with any of the displays
you have, but even then I don't think more kernel support is needed for this.
After all the userland application only has to open the right device in /dev
and do whatever it wants with it. Kernels have been able to handle multiple
ports for a very long time, things are different for multiple graphic cards.

What I'm talking about is moving the protocol decoding part out of the
user-land applications so that all of them don't have to reinvent the wheel
every time they want to do event handling outside of X11.



---
Stephane Peter                          megastep@lokigames.com
Programmer
Loki Entertainment Software