Array of strings not displayed in Variable Explorer pane

[jitseniesen]

The array

np.array([[a, b],
          [c, d]])

is displayed as ndarray object of numpy module in the variable explorer (red) instead of listing the values, as would happen for an array of numbers.

Originally posted by @PhilipYip1988 in #22466 (comment)

[PhilipYip1988]

I think the memory of a ndarray of text is preallocated from the supplied text data, with 4 bytes used for each Unicode character:

>>> import numpy as np
>>> nums = np.array([[1, 2], [3, 4]])
>>> nums.dtype
dtype('int64')
>>> arr1 = np.array([['a', 'b'],['c' ,'d']])
>>> arr1.dtype
dtype('<U1')
>>> arr5 = np.array([['abcde', 'b'],['c' ,'d']])
>>> arr5.dtype
dtype('<U5')

>>> arr1[0, 0] = 'yellow'
>>> arr1[0, 0]
np.str_('y')

>>> arr5[0, 0] = 'yellow'
>>> arr5[0, 0]
np.str_('yello')

Currently the type shown in Spyder is Array of str32 and Array of str160, it might be clearer if Array of <U1 and Array of <U5 is used, more consistent to the dtype attribute:

There is slightly inconsistent behaviour when indexing into a numeric array for example of np.int8 and replacing it with a value that is out of the range of the datatype of the array. I think one major change with numpy 2 was to avoid silent truncation for numeric data outlined in NumPy 2.0 Migration Guide and NEP50.

However it seems that NEP50 isn't implemented for the text data and a silent truncation is still carried out: