Calibration file overwrites CONSTANTS["AdditionalDropCollectionsREC"]

[dudarboh]

In ILDReconstruction.py writing edm4hepfiles follows this logic:

    edm4hepOutput = PodioOutput("EDM4hepOutput")
    edm4hepOutput.filename = f"{reco_args.outputFileBase}_REC.edm4hep.root"
    edm4hepOutput.outputCommands = ["keep *"]
    for name in CONSTANTS["AdditionalDropCollectionsREC"].split(" "):
        edm4hepOutput.outputCommands.append(f"drop {name}")

So, naively one assumes that to drop additional collections one needs to add to the CONSTANTS definition at the top a corresponding key, e.g:

CONSTANTS = {
    "CMSEnergy": str(reco_args.cmsEnergy),
    "BeamCalCalibrationFactor": str(reco_args.beamCalCalibFactor),
    "AdditionalDropCollectionsREC" : "BCAL" # This new key is added manually to drop BCAL collection
}


det_calib_constants = import_from(f"Calibration/Calibration_{det_model}.cfg").CONSTANTS
CONSTANTS.update(det_calib_constants)


parseConstants(CONSTANTS)

However, this does not drop the BCAL collection because CONSTANTS.update(det_calib_constants) overwrites this key with a value from the calibration file.

Thus, one must define AdditionalDropCollectionsREC only after the calibration file is read, which overwrites whatever is dropped inside calibration files.

[tmadlener]

Thanks for the report. This is indeed unintentional. This most likely also affects the LCIO output, I would assume(?). The way I see it there are different fixes possible

• Attach more collections to AdditionalDropCollectionsREC only after the CONSTANTS have been loaded (and parsed) as you have proposed it.
• Add a different key to the top-level CONSTANTS that doesn't get overwritten and use that in addition to the AdditionalDropCollectionsREC when filling the edm4hepOutput.outputCommands.

I would have a slight preference for the second, but you are potentially quicker with fixing this than me, so "first come first merge" (or something along those lines).