Input device design (3)

[Jim Gettys]

On Wed, 2005-08-31 at 20:58 -0400, James Cloos wrote:
> >>>>> "Joe" == Joe Krahn <krahn at niehs.nih.gov> writes:
> 
> Joe> ... event packets can be packaged to to the same event data
> Joe> formats used by X clients. The code is all there, and will also
> Joe> be there when new event types are defined. ... It is one more
> Joe> IPC exchange at most. An external low-load I/O daemon handling
> Joe> SIGIO from many devices might even be a good thing.
> 
> So, which is better:  moving the input event code into a separate
> daemon or just moving it into a(n optionally shared) library the
> x server can link to?

I claim the latter: process scheduling on some systems is an uncertain
thing, and people really don't like latency variation.
				- Jim