Allen hangs on very large events
While trying to run over some PbPb MC using the standalone CPU build (mdf at ), I noticed that Allen hangs on very large events. I was able to track the issue to the MDFProvider (https://gitlab.cern.ch/lhcb/Allen/-/blob/master/main/include/MDFProvider.h) and transpose_event (https://gitlab.cern.ch/lhcb/Allen/-/blob/master/main/src/Transpose.cpp). It seems like Allen waits for a slice large enough to hold the event and just waits forever if no slice is large enough. In this case, I would've expected the event to just be passed through using @dcraik's mechanism for handling large events, but I'm not sure it covers this case. I was able to get around it by increasing the hard-coded minimum slice size here: https://gitlab.cern.ch/lhcb/Allen/-/blob/master/main/include/MDFProvider.h#L180. I'm not sure what an actual fix would be, but I think this case should at least fail in a useful way. There's probably also a similar issue with the MEPProvider.
You can find a small PbPb MDF at /afs/cern.ch/user/t/thboettc/public/upgrade-202004-PbPb-EPOS-b-6_14fm. The second event in the file is about 4MB and should produce the hanging behavior.