Repository navigation
Mask from MaskedColumn is not copied #7362
Description
When copying a masked table, the mask of MaskedColumn objects is not copied, which leads to weird result:
In [1]: %astropy
Numpy 1.14.2
Astropy 3.1.dev21681
In [2]: t = Table([[2,4,1,3,6], [3.,2.,np.nan,4,6]], names=('a', 'b'), masked=True)
In [3]: t['b'] = np.ma.masked_invalid(t['b'])
In [4]: t
Out[4]:
<Table masked=True length=5>
a b
int64 float64
----- -------
2 3.0
4 2.0
1 --
3 4.0
6 6.0
In [5]: t2 = t.copy()
In [6]: t2.sort('a')
In [7]: t2
Out[7]:
<Table masked=True length=5>
a b
int64 float64
----- -------
1 --
2 3.0
3 4.0
4 2.0
6 6.0
In [8]: t
Out[8]:
<Table masked=True length=5>
a b
int64 float64
----- -------
2 --
4 2.0
1 nan
3 4.0
6 6.0
Looking at https://github.com/astropy/astropy/blob/master/astropy/table/column.py#L1122-L1133, It seems that the data is copied but not the mask. Maybe @mhvk ?
Activity
yannick1974 commentedon Apr 11, 2018
Note that the problem exists when using numpy 1.14.2 but not numpy 1.13.3. Probably linked to https://docs.scipy.org/doc/numpy/release.html#compatibility-notes
saimn commentedon Apr 11, 2018
Yes indeed, it is due to the fact that the mask is no more always copied (related to #5060, #5530).
Fix copy of mask when copying a Table
Fix copy of mask when copying a Table