Please test against astropy v5.2rc1 and report back

Hello. astropy v5.2rc1 is out. Please test against it and report back success or failure before v5.2 is really released. Thank you for your help!

[barentsen]

But just to double check, does this use the new “cloud FITS” feature that you added to astropy recently?

At this time Lightkurve only uses the MAST TessCut API via Astroquery, so it does not directly depend on “cloud FITS” yet.

(But it could be a great performance improvement in the future!)

[bmorris3]

Not sure if this is related or not, but using astropy 5.2 and lightkurve 2.3, I’m trying to do a simple median on the flux:

from lightkurve import search_lightcurve

name = "KIC 8006161"

light_curve = search_lightcurve(
    name, mission='Kepler',
    cadence='short'
).download_all()
np.median(light_curve[0].flux)

which returns

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Cell In [4], line 1
----> 1 np.median(light_curve[0].flux)

File <__array_function__ internals>:200, in median(*args, **kwargs)

File ~/miniconda3/envs/research/lib/python3.9/site-packages/astropy/units/quantity.py:1818, in Quantity.__array_function__(self, function, types, args, kwargs)
   1805 # A function should be in one of the following sets or dicts:
   1806 # 1. SUBCLASS_SAFE_FUNCTIONS (set), if the numpy implementation
   1807 #    supports Quantity; we pass on to ndarray.__array_function__.
   (...)
   1815 # function is in none of the above, we simply call the numpy
   1816 # implementation.
   1817 if function in SUBCLASS_SAFE_FUNCTIONS:
-> 1818     return super().__array_function__(function, types, args, kwargs)
   1820 elif function in FUNCTION_HELPERS:
   1821     function_helper = FUNCTION_HELPERS[function]

File ~/miniconda3/envs/research/lib/python3.9/site-packages/astropy/utils/masked/core.py:864, in MaskedNDArray.__array_function__(self, function, types, args, kwargs)
    862 dispatched_function = DISPATCHED_FUNCTIONS[function]
    863 try:
--> 864     dispatched_result = dispatched_function(*args, **kwargs)
    865 except NotImplementedError:
    866     return self._not_implemented_or_raise(function, types)

File ~/miniconda3/envs/research/lib/python3.9/site-packages/astropy/utils/masked/function_helpers.py:592, in median(a, axis, out, **kwargs)
    590 if NUMPY_LT_1_25:
    591     keepdims = kwargs.pop("keepdims", False)
--> 592     r, k = np.lib.function_base._ureduce(
    593         a, func=_masked_median, axis=axis, out=out, **kwargs
    594     )
    595     return (r.reshape(k) if keepdims else r) if out is None else out
    597 else:

File ~/miniconda3/envs/research/lib/python3.9/site-packages/astropy/units/quantity.py:1255, in Quantity.__iter__(self)
   1253 def __iter__(self):
   1254     if self.isscalar:
-> 1255         raise TypeError(
   1256             f"'{self.__class__.__name__}' object with a scalar value is not"
   1257             " iterable"
   1258         )
   1260     # Otherwise return a generator
   1261     def quantity_iter():

TypeError: 'MaskedQuantity' object with a scalar value is not iterable

This can be avoided if you intentionally dodge the astropy.utils.masked.Masked object like this:

np.nanmedian(light_curve[0].flux.unmasked)

which returns the expected result.