monotonically increasing check should be ">="

[greglucas]

A monotonic function can have values that are equal to each other
https://en.wikipedia.org/wiki/Monotonic_function

so I believe this should be >=

cdflib/cdflib/xarray/xarray_to_cdf.py

Line 471 in 88ba11e

return np.all(epoch_data[1:] > epoch_data[:-1])

It does seem like there is some ambiguity about whether monotonically increasing includes the equal sign or not. I couldn't figure out which definition ISTP compliance would want. "strictly increasing" would be how I would interpret the code currently.

[ericthewizard]

A time series should be increasing in time (with no duplicate times); "strictly increasing" is probably better wording.

There's definitely some code out there (e.g., in IDL SPEDAS) that uses this assumption for performance reasons (I vaguely remember coming across it in the past).

[bryan-harter]

So if we are being fully ISTP compliant, we want to keep the check as ">" rather than ">="?

[ericthewizard]

So if we are being fully ISTP compliant, we want to keep the check as ">" rather than ">="?

I don't know if it's an ISTP thing (ISTP compliance is mostly for the metadata, rather than the data/time values themselves), but I know a lot of systems will break if there are duplicate timestamps. Even if a system is built to support duplicate timestamps, it brings up the questions: which data value is correct for the timestamp? and if they're the same, why include the duplicates?

No idea if the SPDF metadata editor checks this at all when running an ISTP compliance check, but I think it's a good thing to check somewhere (though places where this shows up tends to be at file boundaries, so compliance checks would miss it without checking multiple files at once).