When there are no device files found, jinput won't try to read the 0th
one.
When no js* devices are found in /dev/input it will look in /dev
Try to free some of the memory created in the init process
doesn't happen properly and events are left of the ff device meaning we
run out of device space. Added a thread that init and cleanup are called
from via methods on the interface.
Then it became apparent that rumble would only work from the same thread
as the others too, so thats added in there.
Nastyness all round.
java.home set to the jre dir, even though a 'which java' reveals it's
using the jdk one, not the jre one. This causes the include dir not to
be found and causes the build to fail. Made it so that the build file
looks for includes relative to the jre dir and the jdk dir, so all
systems should work (yeah, right)
Sigsgv with logitec USB keyboard with volume controls and mouse wheel
when valid event device list in /dev/input/event* has invalid devices in the middle of the sequence, a sigsegv is seen
statements back in every time something breaks
Fixed a crash under linux if when it was thinking there was at least one
more joystick that it really had, funnily enough when it tried to get
data from this joystick, it all went bang.
