There was a regression in the mqueue transport from https://github.com/orocos-toolchain/rtt/commit/a34a19ad740fc15a81574ade66c5266fdf340576. The inputReady() call on the receiving side of the channel did not block anymore on the data sample being received and therefore the corba_mqueue_ipc_test can fail depending on the number of CPUs available and load.
Furthermore I improved both tests that involve IPC, the corba_ipc_test and corba_mqueue_ipc_test to be more resilient against execution under high CPU load or on slow machines. Additionally, a resetCallBackPeer() method has been introduced in order to allow multiple executions of the test (the client) against the same running server process.
There was a regression in the mqueue transport from https://github.com/orocos-toolchain/rtt/commit/a34a19ad740fc15a81574ade66c5266fdf340576. The inputReady() call on the receiving side of the channel did not block anymore on the data sample being received and therefore the
corba_mqueue_ipc_testcan fail depending on the number of CPUs available and load.Furthermore I improved both tests that involve IPC, the
corba_ipc_testandcorba_mqueue_ipc_testto be more resilient against execution under high CPU load or on slow machines. Additionally, aresetCallBackPeer()method has been introduced in order to allow multiple executions of the test (the client) against the same running server process.This PR will hopefully fix the test errors on Travis for the toolchain-2.9 branch.