Use Lifetime::QA for sampled quantities

[ktf]

This will make sure that sampled data will not be used to do counting
of in-fly timeframes.

[davidrohr]

Is it safe to just overwrite it? Shouldn't we request the developer to set it, and give a failure if not? Otherwise I think this can be confusing?

[ktf]

@knopers8 @Barthelemy in order to correctly account for how many timeframes have been processed, we need to have all OutputSpec which are associated to sampled data marked as Lifetime::QA. While this will not change how the data is processed, it will make sure it will not be accounted to consider a timeframe "complete" (because only a fraction of the timeframes have it, by definition).

Do you see any problems with doing it?

[ktf]

@davidrohr see the other message. It Lifetime::Timeframe and Lifetime::QA should be identical in terms of processing, QA simply means that the data is sampled or created once in a while (like in the case of Analysis histograms).

[davidrohr]

Then why do we call it Lifetime::QA? Can we rename it to Lifetime::Sampled or something like that?

[ktf]

That's what I called it historically at the dawn of DPL. I am happy to rename it, however I do not like "Sampled" much, since it needs to apply also for histograms, which fit "Accumulated" more than "Sampled". Maybe "Occasional"?

[davidrohr]

In some sense I would call it Optional, but we have optional already for another purpose. What is actually exactly the difference?
How about sporadic / irregular / arbitrary / selective ?

[Barthelemy]

Hi,
@knopers8 is absent till next Monday and I think that he is the one who should give his green light. I propose to wait for him.

[davidrohr]

Hi @Barthelemy : this is actually on the critical path for commissioning, so I would not wait till Monday. Worst case we can revert. But it seems this breaks the DataSampling ctest. That must obviously be fixed.

[ktf]

Are you sure you are on the right branch?

[Barthelemy]

Hi @Barthelemy : this is actually on the critical path for commissioning, so I would not wait till Monday. Worst case we can revert. But it seems this breaks the DataSampling ctest. That must obviously be fixed.

Ok, I think that I fixed some of the failing tests. Although, @knopers8 will have to review it when he is back.

However for lines 112 and 113 I don't know how to fix it. If I put a lifetime, then the subspec is wrong, if I don't put it, it fails.

[ktf]

I fixed the remaining tests.

[davidrohr]

The test is still failing:

 58/401 Test  #67: test_DataSampling_test_DataSampling ..................................................***Failed    0.14 sec

[Barthelemy]

ok, forgot this one.

[davidrohr]

ok, forgot this one.

and it is not the DDS test :)

[Barthelemy]

hopefully it is ok now

[ktf]

So it seems ok now. @Barthelemy Can we merge?

[knopers8]

I think this should be fine, but I will immediately contact you if I see something not working anymore.