Describe the issue:

Description of the Error

I encountered an error when having a MaskedArray with dtype set as StringDType and trying to set a fill_value other than None (or using the filled(fill_value) method).

The error seems to come from a data type check of the fill_value before setting it to the MaskedArray here:

Possible fix

Currently it is just admitting OSVU as valid numpy.dtype.char types, when in https://numpy.org/doc/stable/reference/generated/numpy.dtype.kind.html
the T character code for StringDType could be admissible.

I tested rudimentarily that including T in OSVU for the check works just fine.

Reproduce the code example:

# Here the masked array is not created because `fill_value=''` (or any other string) is not accepted.

import numpy as np
from numpy.dtype import StringDType

strdt_ma = np.ma.MaskedArray(['zero', 'one', 'two', '', 'four'], mask=[False, False, False, True, False], fill_value='', dtype=StringDType(na_object='', coerce=True))


# Here the masked array is created successfully because `fill_value = None`, but fails when an string value is used

strdt_ma = np.ma.MaskedArray(['zero', 'one', 'two', '', 'four'], mask=[False, False, False, True, False], fill_value=None, dtype=StringDType(na_object='', coerce=True))

print(strdt_ma.dtype,strdt_ma.dtype.char)
strdt_ma.fill_value = '' # fails for any string
print(strdt_ma.filled('N/A')) # fails for any string
print(strdt_ma.filled()) # the missing value is cast as '?'
strdt_ma.filled()[3] == None # False
strdt_ma.filled()[3] =='?' # True

Error message:

Traceback (most recent call last)
Cell In[242], line 1
----> 1 strdt_ma = np.ma.MaskedArray(['zero', 'one', 'two', '', 'four'], mask=[False, False, False, True, False], fill_value='', dtype=StringDType(na_object='', coerce=True))

File /opt/anaconda3/envs/py313/lib/python3.13/site-packages/numpy/ma/core.py:3017, in MaskedArray.__new__(cls, data, mask, dtype, copy, subok, ndmin, fill_value, keep_mask, hard_mask, shrink, order)
   3015 # But don't run the check unless we have something to check.
   3016 if fill_value is not None:
-> 3017     _data._fill_value = _check_fill_value(fill_value, _data.dtype)
   3018 # Process extra options ..
   3019 if hard_mask is None:

File /opt/anaconda3/envs/py313/lib/python3.13/site-packages/numpy/ma/core.py:504, in _check_fill_value(fill_value, ndtype)
    501 elif isinstance(fill_value, str) and (ndtype.char not in 'OSVU'):
    502     # Note this check doesn't work if fill_value is not a scalar
    503     err_msg = "Cannot set fill value of string with array of dtype %s"
--> 504     raise TypeError(err_msg % ndtype)
    505 else:
    506     # In case we want to convert 1e20 to int.
    507     # Also in case of converting string arrays.
    508     try:

TypeError: Cannot set fill value of string with array of dtype StringDType(na_object='')

Python and NumPy Versions:

import sys, numpy; print(numpy.__version__); print(sys.version)

2.3.1
3.13.5 | packaged by Anaconda, Inc. | (main, Jun 12 2025, 11:23:37) [Clang 14.0.6 ]

Runtime Environment:

No response

Context for the issue:

Recently I started to work with numpy>=2 to be able to use the functionality of variable width string arrays, and in so avoid unwanted string truncation during manipulation. I ran into this issue when trying to use MaskedArray with the StringDType and noticed the restriction in the fill_value definition.

In conclusion, the issue should be solved by adding T dtype character code to the accepted list in _check_fill_value() function.

PD: I'm a long-time user and first time contributor, I don't mind submitting the change myself following the contributor guidelines https://numpy.org/devdocs/dev/index.html#development-process-summary but I'm not sure if there is anything else to do to be allowed to contribute. Thanks.