Open belforte opened 7 years ago
Hi Stefano,
The reason why the output dataset name appears in the json content of that file is because you can have more than one output dataset in a task. So the content of the file is: {"output dataset name 1": LumiList 1, "output dataset name 2": LumiList 2, etc} Of course you can put separate LumiLists in separate files. Evidently I preferred to put them all in one file. Note that the name of the file and the crab report output clearly refer to output dataset"S". So, I would say it is not a bug, but a feature :-)
Best, Andres
I suspect it was meant to be a feature. Original developer thought that maybe we want to publish multiple datasets, so the outputDatasetsLumis.json file is built like a list of dictionaries, each having a dataset name as key and the lumilist as value. Of course this was not documented. Examples in the twiki do not mention this file: https://twiki.cern.ch/twiki/bin/view/CMSPublic/CRAB3FAQ#Recovery_task https://twiki.cern.ch/twiki/bin/view/CMSPublic/CRAB3FAQ#Dealing_with_a_growing_input_dat
I guess it should be addressed as a "Fix documentation". And worry later if code should be changed to be more what one would expect.
OK. We agree ! Thanks @AndresTanasijczuk
Btw, the fact that crab report produces this file is mentioned in https://twiki.cern.ch/twiki/bin/view/CMSPublic/CRAB3Commands#crab_report
see: https://github.com/dmwm/WMCore/issues/7537
need to check https://github.com/dmwm/CRABClient/blob/master/src/python/CRABClient/Commands/report.py