[PD] [hid] steady data stream vs. output on event

[Hans-Christoph Steiner]

I am diving back into working on [hid], I am porting it to Windows  
thanks to sponsorship from http://vibrez.net.  There is a lingering  
question which I would like to address, so I am looking for feedback  
from this list.

Right now, [hid] basically outputs events in response to the OS.  It  
polls the event queue at a given frequency and processes everything in  
the queue on each cycle.  This means that the data stream is a bit  
erratic.

Here's how I am currently thinking of handling this:

- for buttons/keys, process all events in the queue

- For absolute axes, output the most recent event per poll, since  
absolute axes are position, and its not really necessary to know how  
many events there were between [hid] polls, just the position where it  
ended up.  Then it would output one event per poll cycle.

- relative data is the trickiest, mouse for example.  The data is in  
effect the velocity.  A mouse tells you how many units it has travelled  
per refresh cycle.  But when the mouse stops moving, it doesn't not  
send a "0" velocity.  it could be made more like velocity and have it  
polled for the current velocity.  This would give a steady stream of  
data, but then it would not give accurate info about how far the mouse  
as travelled.  Perhaps, [hid] would just output everything it finds in  
the queue, then I write a separate object which gives the average  
velocity per poll period, based on the distance in all events it  
receives.

The first two apply directly to sensor data, I think, I am not sure how  
relative axes with sensors compare to mice.

.hc

________________________________________________________________________ 
____

"Computer science is no more related to the computer than astronomy is  
related to the telescope."
                                                           -Edsger  
Dykstra