2009/1/28 Andy Green <[email protected]>: > If you want to improve that, feel free to send a patch.
I'm not really sure what the right approach would be, because i have never seen a kernel option that allows you to enter a list of operations to perform. If you have any hints, feel free to drop me a note! > Why service hardware requirements based off an ISR in userspace when it > can be decimated inside the ISR? It's that which seems ass-backwards. Is this really the case? I mean IMO the job of the kernel is to interface hardware and to do things that can't be done in userspace due to performance or hardware access reasons. So in my point of view, the kernel should just send out the unfiltered evens and let userspace decide what to do with it. And if userspace decides to do finger-tracking via bayesian tracker, it is ok! But again, this is my opinion and I also accept your (Nelson, Andy, Werner, etc) view. As Werner said later in the thread: all arguments are written down... So let's look forward and find a better solution to the "compose a custom filterchain" problem! Sven
