This addresses issue #196 by fixing the conversion of the timeout in milliseconds to the struct timeval in libgearman-server/connection.cc. This seems the simplest way to fix this rather egregious bug. I still think seconds is the more intuitive unit for a job timeout, but that would require more widespread changes and I haven't received much feedback preferring that as a solution.
Any objection to my modification of the PROTOCOL file to firmly document that the timeout should be specified in milliseconds?
This addresses issue #196 by fixing the conversion of the timeout in milliseconds to the
struct timevalinlibgearman-server/connection.cc. This seems the simplest way to fix this rather egregious bug. I still think seconds is the more intuitive unit for a job timeout, but that would require more widespread changes and I haven't received much feedback preferring that as a solution.Any objection to my modification of the PROTOCOL file to firmly document that the timeout should be specified in milliseconds?