[log in to unmask]">



I am seeing another strange issue when testing the monitoring application. 


I use a Driver that runs the track reconstruction on each event, and it seems to work fine.  But when I disconnect my monitoring session and reconnect, it seems to be break things.  Basically after reconnecting I see the plots but only the RawTrackerHit collection has any hits, and subsequent collections from reconstruction like TrackerHits are always empty for every event.  LCSim is actually setup from scratch between each monitoring session, including the Detector and conditions.  What I suspect here is that there is some static class or data structure used in the track recon which is not being properly replaced or refreshed for the new run.  It might for instance contain stale references to old objects on the Detector from the previous session used as map keys.  Obviously this would break the reconstruction if these objects do not match those being currently used.


I’m trying to rule out problems with the monitoring app itself, and I have mostly convinced myself that this is not something caused by that code, as the RawTrackerHit plots still look fine.


Has anyone run into problems like this?  Any idea on how to fix it?


(In general, I think we need to take a look at the use of static methods and classes in the reconstruction, as it can easily lead to issues such as this.)



Use REPLY-ALL to reply to list

To unsubscribe from the HPS-SOFTWARE list, click the following link: