Fix unit processing for percent and permille

[theHenks]

Parse o/o and o/oo as percent and permille units while loading. These units are used in the context of the FlashCam decoding in the extra section (s. here)

Why need ASAP a description of units in the dataformat-specs. This is way too much work to debug every time.......

c.f. @gipert @ggmarshall

@DaGeibl : I still need to update the tests/ on LegendHDF5IO.jl and the bump the compat for this too work properly.

[gipert]

well these are also not physical units... otherwise we also accept ADC

[theHenks]

well these are also not physical units... otherwise we also accept ADC

However, we can parse them into Unitful The unit percent and permille exist. The question is if we wanna drop them in the future or not.

[oschulz]

I see little value in keeping such units around, I would definitely prefer stripping them. What would be the use case for such "units"?

[theHenks]

I see little value in keeping such units around, I would definitely prefer stripping them. What would be the use case for such "units"?

We can also strip them. However, you loose now information that the numbers are permille. At least in the context of JuLeAna, we save so far percent etc as units, e.g., in the pars output for survival fractions etc.

[oschulz]

We can also strip them. However, you loose now information that the numbers are permille

Ah - no, with stripping I mean converting them to a normal fraction value.

[oschulz]

well these are also not physical units... otherwise we also accept ADC

It's not quite the same - ADC is not a universally comparable thing, percent is well-define. Technically, it's not a unit, but a mathematical constant, but for practical purposes one can treat it as kind of a unit.

[oschulz]

But why o/o and not % and o/oo and not ‰?

[theHenks]

But why o/o and not % and o/oo and not ‰?

I don't know. But this can be changed in the link I send. I think this all comes down again to a dataformat spec concerning units...

[theHenks]

well these are also not physical units... otherwise we also accept ADC

It's not quite the same - ADC is not a universally comparable thing, percent is well-define. Technically, it's not a unit, but a mathematical constant, but for practical purposes one can treat it as kind of a unit.

As I said, Unitful is treating it as a unit.

[oschulz]

I don't know. But this can be changed in the link I send. I think this all comes down again to a dataformat spec concerning units

Is this mainly intended for output that's both human and machines-readable, though? Do we already use it somewhere except for survival fractions in pars?

[theHenks]

I don't know. But this can be changed in the link I send. I think this all comes down again to a dataformat spec concerning units

Is this mainly intended for output that's both human and machines-readable, though? Do we already use it somewhere except for survival fractions in pars?

Dont know, maybe @gipert can answer.

[gipert]

I think it's pretty useless to store values in percent or permill...

what punishment should we inflict to @ssailer?

[ssailer]

Please discuss this with the Physics King. I was told, we're supposed to make identical copies of the daq files.

[theHenks]

Please discuss this with the Physics King. I was told, we're supposed to make identical copies of the daq files.

Is that @ggmarshall ? I guess whatever we do we probably won't reprocess all daq files again to change the unit in these couple keys, so we have to fix it here independent of that.
My proposal:

• Let's drop the units here and get rid of them followed by updating LegendHDF5IO (and the tests) and check the testdata (someone from pygama team)
• PR to legend-dataformat-specs with a unit definition. All units according to the SI unit conventions, no percent, permille, ADC etc.