On 21/12/2011 01:52, Weijun Wang wrote:
I think this timeout is specified by elapsed but the implementations use JVM_CurrentTimeMillis or in the case of Windows, a timeout provided by the kernel. My guess is that if this test is changed to use System.currentTimeMillis that it may be more robust but you will need to try it.:I'm not sure. The time consumed is from Socket::setSoTimeout(). Do you know if it's connected to clock time (currentTimeMillis) or elapsed time (nanoTime)?
-Alan
