[ANNOUNCE] xf86-input-fpit 1.4.0

[Alan Cox]

> the "standard" is pretty much defined by what the driver can take. If it
> can't parse the protocol then the device is rather useless anyway.
> but really, writing a serial kernel driver is rather trivial and has a
> higher chance of actually working long-term than dragging the old input
> drivers along.

Except that you need to write a lot of input drivers - for Linux, for
OpenBSD, for NetBSD, for Dragonfly etc..

In the Linux case yes you could write an fpit to Linux input layer ldisc
but that only fixes the world for a single OS.

Re; polling - a lot of these fpit devices have weird IRQs for the serial,
you may find they are not in fact polled but need a suitable IRQ forcing.