Problems with stopwatch test

[David Jencks]

I'm having inconsistent problems with the new geronimo commons 
StopWatch test on my slow old powerbook.  Perhaps in an attempt to get 
on to more important things, apparently sometimes sleep(1000) takes 
only 999 milliseconds.

I'm curious if anyone else observes anything similar and if anyone has 
a suggestion about how to work around this. (buying a new computer is 
not practical at the moment, unfortunately).  I patched the test to 
show some instances of this problem.

Thanks
david jencks