Nkx111 patch 4

[nkx111]

Some bug fix and new features:

1. Fix bugs when opening files with unknown event types(e.g., v2.2 file). Previsuly in this case the input event will be initialized with data members, but the method cannot be called, and causes seg.fault.
2. added method: TRestEventProcess::GetParallelDataMembers()
3. restRoot macro loading now also working on windows.
4. added validation pipeline to test open v2.2 files.
5. fixed Console::kbhit() on windows.

[jgalan]

Hi @nkx, perhaps we could create a dedicated data repository, where we place data files such as the one you are adding v2.2.30_hits.root?

This would be something similar to https://github.com/rest-for-physics/axionlib-data/tree/36629372bcffddd6f95eb12a565c1b15475a64be

That we later make http accessible, and use at the pipelines using wget.

[jgalan]

What the GetParallelDataMembers is doing?

[juanangp]

Where is this template used?

[juanangp]

Personally, I am not fan of macros. Also, I cannot see where this macro is used.

[nkx111]

It could used in for example in TRestRawSignalAnalysisProcess, to replace the cut code.

if (ApplyCut()) return nullptr; return fSignalEvent; --> _ApplyCut(fSignalEvent)

I think it is more clear to use this macro.

[nkx111]

Hi @nkx, perhaps we could create a dedicated data repository, where we place data files such as the one you are adding v2.2.30_hits.root?

This would be something similar to https://github.com/rest-for-physics/axionlib-data/tree/36629372bcffddd6f95eb12a565c1b15475a64be

That we later make http accessible, and use at the pipelines using wget.

It sounds interesting. We also have several root files used in pipeline for process input, and generator files for NLDBD simulation. Those files are much larger than others.

[nkx111]

What the GetParallelDataMembers is doing?

This method is used in another library repository which is not open-sourced yet. It can be useful when you have objects to fill during the process and to summarize after processing. For example we have a process that accumulates value A and B, and calculate C=A/B at the end. Then, in case of multi-thread environment, A, B and C are splited in to several parts, the calculation could be wrong. By using this method, in thread 0, we can retrieve A(B) from thread 1,2,3..., and merge them manually. Then we can correctly calculate C.