Moved exceptions from init and added debug statements to process #1565
Conversation
|
Capybara summary for PR 1565 |
| // Check if the algorithm has been initialized properly | ||
| if(!m_initialized){ | ||
| debug("Initialization did not complete"); | ||
| return; |
There was a problem hiding this comment.
Unless you throw here, the Process will be called over and over.
| return; | |
| throw std::runtime_exception("Initialization did not complete"); |
There was a problem hiding this comment.
Will it not call process over and over anyway? I thought the behaviour we wanted was to flag the error but still return an empty collection so that the reconstruction can be carried out with the limited geometry provided. It would be nicer if upon throwing the exception in init, JANA knows not to try again.
There was a problem hiding this comment.
It will not. The exceptions in Process trickle into our JEventProcessorPODIO, where we try to print error messages only once, which is the desired effect. We could do some accounting to disable collections, but, you are right, they don't get disabled from being called currently.
|
This solves a global issue locally. We should figure out why exceptions in |
Briefly, what does this PR introduce?
Highlighted by @ShujieL and @wdconinc https://chat.epic-eic.org/main/pl/uiei8z865tbdjn4mpou19o4tuo. Failing to initialise the FarDetectorTrackerCluster algorithm, due to missing detectors in the geometry, calls an exception meaning JANA tries to init it again on the next event resulting in errors printed for each event.
This PR tries to tidy up the output by removing the exception so that the initialisation error is only printed once. A check is instead moved to the start of process() where repeated warning messages can still be accessed by setting the logger to debug.
What kind of change does this PR introduce?
Please check if this PR fulfills the following:
Does this PR introduce breaking changes? What changes might users need to make to their code?
NA
Does this PR change default behavior?
Fewer repeated error statements should be printed. Otherwise no.