covers_positions routine with diverging return types

[poplarShift]

opendrift/opendrift/readers/operators/readerops.py

Line 46 in 857fb75

def covers_positions(self, lon, lat):

returns an array

whereas these, and all the other ones I can find, return a boolean:

opendrift/opendrift/readers/basereader/unstructured.py

Line 134 in 857fb75

def covers_positions(self, x, y, z=0):

opendrift/opendrift/readers/interpolation/structured.py

Line 143 in 857fb75

def covers_positions(self, x, y, z=None):

what is the reasoning behind that?

[gauteh]

That seems to be a mistake. The latter two are probably the correct ones, but I am a bit surprised they don't return arrays of booleans.

[knutfrode]

I am not sure if there is a real reason for this diverging practice. I will have a look.

[gauteh]

I don't think the one in readerops have ever been used in practice before.

[knutfrode]

The methods actually belong to two different classes:
readers/interpolation/structured.py is asubclass of ReaderBlock whereas the others are subclasses of BaseReader

But anyway, I agree with Gaute that it sounds more logical for all of these methods to return an array of booleans.
Then we could have a simple wrapper onliner method covers_all_positions the simply returns True of all booleans are True.

If you agree, I could make this change consistently.

[gauteh]

Yes, that sounds good.

For ReaderBlock it should have a different name: covers_all_positions, while for the readers it should return indices?

[poplarShift]

Thanks! Does this have any bearing on my latest comment #1279 about the possible return of a zero size array?