tuple.__str__ call __repr__ and not __str__ for underlying items.

[ccossou]

Bug report

Bug description:

After numpy 2.0, repr changed for all numpy types (e.g; np.int64 and np.float64).

This had an impact on tuple.str, because it seems that tuple.str refers to item.repr instead of str.

Here, I use Python 3.12.8 and numpy 2.2.0

We now have this behavior:

>>> import numpy as np
>>> x = np.int64(45)
>>> print((x,x))
(np.int64(45), np.int64(45))
>>> print(x)
45

If you use python 3.11.11 and numpy 1.26.0, you had:

>>> import numpy as np
>>> x = np.int64(45)
>>> print((x,x))
(45, 45)
>>> print(x)
45

I personally think that the type has no place in a str conversion, but that's my opinion. Thus, I'm not sure it's a bug or a feature. But I would tend to think that tuple.str should call str and tuple.repr should call repr

CPython versions tested on:

3.11, 3.12

Operating systems tested on:

Linux

[picnixz]

It would be a feature and a breaking-change feature. For most simple containers, the __str__ method is the same as their __repr__ method (this is already the case in Python 3.9). The docs also say (emphasis mine):

Called by str(object), the default __format__() implementation, and the built-in function print(), to compute the "informal" or nicely printable string representation of an object. The return value must be a str object.

This method differs from object.__repr__() in that there is no expectation that __str__() return a valid Python expression: a more convenient or concise representation can be used.

Now, if you want to make containers use __str__ for their items, this should be discussed on Discourse (https://discuss.python.org/c/ideas/6) and likely would need a PEP. I'm going to close this one as not planned (sorry, this issue requires a wider discussion).