Refactor II

[Zehvogel]

Next batch in the refactoring series. Mostly just moving (almost) all the definitions from the main steering file to sub files and then loading them with the sequenceLoader copied from iLCSoft/ILDConfig#137. The latter can be removed once we have it (or something similar) in key4hep proper.

BEGINRELEASENOTES

• Move most parts of the reconstruction steering into sub files loaded by the sequenceLoader

ENDRELEASENOTES

There will be a third batch of patches to add some more configuration flags and refactor the output definitions soon ™️

[Zehvogel]

@tmadlener can you consent to the py_utils.py I stole from you being included here under apache2 instead of gpl3 as in ILDConfig? :)

[tmadlener]

@tmadlener can you consent to the py_utils.py I stole from you being included here under apache2 instead of gpl3 as in ILDConfig? :)

Sure, no problem :) As I have already alluded to in the original pull request for k4FWCore, I am not yet entirely happy with the SequenceLoader I came up with because it has a few too many implicit assumptions. Once, I have a few more spare cycles I would like to do a proper AlgorithmList or something in that direction with similar functionality. However, from the SequenceLoader to that the refactoring should be minimal (and only be necessary for the "main" config and not the individual modules).

[Zehvogel]

I am not sure why the edm4hep input test does not fail on my machine so it is a bit ugly to debug. It looks related to the PID stuff... @tmadlener maybe you have a good guess what happens? :)

[tmadlener]

PodioOutput ERROR Could not write event, because the following collections are not present: RefinedVertexJets_PID_RefinedVertex

Looks like some whatever produces the RefinedVertexJets only produces the empty collection (which obviously) will not have a PIDHandler attached, so the conversion then fails to create the correpsonding ParticleIDCollection on the EDM4hep side.

[Zehvogel]

Looks like some whatever produces the RefinedVertexJets only produces the empty collection (which obviously) will not have a PIDHandler attached, so the conversion then fails to create the correpsonding ParticleIDCollection on the EDM4hep side.

That sounds likely, yes. :(
Do you have an idea for a fix?

[tmadlener]

The Processor that produces this has to be fixed. It needs to unconditionally add a PIDHandler to the collection. Currently, it probably just produces the empty collection but doesn't attach that.

[Zehvogel]

Oh great thankfully the issue for this already exists...
lcfiplus/LCFIPlus#69