clint-lawrence
commented
4 years ago Some more specific details on the issue here. TwoEdgeSeparation.read() expects _trigger_thread to have an exec_info attribute. However that was removed in #112. My understanding is that this whole contraption with thread is to prevent a test script locking up in the read call in the case that the two edge separation is never actually triggered.
As well as that, it is often the case that you need to do something like
- get the DUT into the correct state to start a test
- prepare the two edge sep measurement (i.e. get it ready to trigger)
- send some stimulus to the jig/DUT to start the event being measured
- then either:
- do a read with a timeout or
- be about the poll the counter to see if the trigger/capture is complete. And then only read if the event was captured.
Hopefully that provides a bit more context to what the concern is here.
The daqmx driver's TwoEdgeSeparation possibly relied upon some of the nastier aspects of the old version of ExcThread (present in 0.5.2). This most likely will need to be reviewed for robustness under error. However we will need to setup some hardware to sufficiently test.