Minor fix to OpHitFinderDeco_module.

[vpec0]

OpHitAlg_deco was trying to use raw::OpDetWaveform timestamp when processing recob::OpWaveform. This is unnecessary as the latter carries the former's timestamp, too. The way the timestamp was retrieved did not give freedom to not have recob::OpWaveforms for some raw::OpDetWaveforms.

[jroto]

Hi @vpec0 , if I understand well, this PR address this issue, right?

[mantheys]

The reason this was implemented like this is the following:
2 TimeStamp classes:
○ raw:OpDetWaveform returns TimeStamp_t which is a double.
○ recob::OpWaveform returns a raw::RDTimeStamp which returns a ULong64_t.
■ Conversion from the original TimeStamp is not possible. Ideas?
■ Possible workaround -> keep using TimeStamp from raw:OpDetWaveform in
OpHitFinder and other modules.

In the deco wvfs, we cannot save negative timestamps. But these are necessary for simulation as the neutrino interaction happens at T0 but the background could produce signals earlier.

[aolivier23]

I'm holding off on reviewing this for now while the discussion continues. I can add it to a release tomorrow if you resolve it by then. Otherwise, I'll be back in a week or so when the next LArSoft release is ready.

[jroto]

Hi @mantheys, please correct me if I'm wrong, but the issues you mentioned were already reported and solved in:
LArSoft/lardataobj#37
LArSoft/lardataobj#40
It looks solved to me, fTimeStamp is now a double in OpWaveform as in OpDetWaveform. This is why I opened this issue last year:
#41
Am I missing something?

[vpec0]

Hi @vpec0 , if I understand well, this PR address this issue, right?

It appears so. I was not aware of this issue, but it now makes sense.

This problem still persists. In the deco wvfs we cannot save negative timestamps. But these are necessary in simulation as the neutrino interaction happens at T0 but background could produce signals earlier. So unless queue find a different solution, I would not apply these changes.

I must agree with Jose here, the TimeStamp() method in recob::OpWaveform now returns double and it is filled with the timestamp of the raw::OpDetWaveform.

[mantheys]

Sorry, my fault! I was not aware of the changes in Larsoft. Happy to see this solved!