Not all float fields should be set to nan by default

[terliuk]

When we compute event_basics, we set all defaults to nan and then overwrite it in processing. In most cases this makes sense, however, it creates issues sometimes. For example, if we have alt_s1 or alt_s2 area as nan, it means that alternatives do not exist and should have area of 0. This might be relevant for some of cut applications or event selection, when no explicit requirement for (non) existence of alternatives is made.

straxen/straxen/plugins/event_processing.py

Lines 293 to 302 in b48fe42

	def set_nan_defaults(buffer):
	"""
	When constructing the dtype, take extra care to set values to
	np.Nan / -1 (for ints) as 0 might have a meaning
	"""
	for field in buffer.dtype.names:
	if np.issubdtype(buffer.dtype[field], np.integer):
	buffer[field][:] = -1
	else:
	buffer[field][:] = np.nan

[JoranAngevaare]

@terliuk this comes up every now and than indeed. It's a question of somehow choosing the defaults such that people don't by accident make wrong selections. This is a very non-trivial question because people will all do something different and/or might not check their data-format properly.

What would you argue makes most sense from your point of view?

[terliuk]

my suggestion would be to use "common sense" (or "physical reality", whatever we name it) for it, so for areas of alternatives we should set 0, for parameters characterizing those non-existing peaks - nans, index -1 etc.