_get_scaling raises TypeError due to misformatted message

[Icyshaman]

Description of your problem

When we are passing tuple (tuple is a valid data-structure for total_size) for total_size as an argument in _get_scaling() we are getting TypeError instead of ValueError even though it should raise a ValueError and this is all because of the use of old-style formatting.

Example:
When we pass a tuple containing more than one Ellipsis it should raise a ValueError but instead of that, it is throwing TypeError.

Please provide a minimal, self-contained, and reproducible example.

_get_scaling((1, 2, 5, Ellipsis, Ellipsis), (2, 3), 2)

Please provide the full traceback.
pymc.distributions.logprob._get_scaling()

Complete error traceback

Input In [1], in _get_scaling(total_size, shape, ndim)
     60     end = total_size[sep + 1 :]
     61     if Ellipsis in end:
     62         raise ValueError(
---> 63             "Double Ellipsis in `total_size` is restricted, got %r" % total_size
     64         )
     65 else:
     66     begin = total_size

TypeError: not all arguments converted during string formatting

Please provide any additional information below.
The above-mentioned issue can be solved by replacing

"Double Ellipsis in `total_size` is restricted, got %r" % total_size

in _get_scaling() by

"Double Ellipsis in `total_size` is restricted, got {}".format(total_size)

Since in present version of code it is trying to unpack the tuple elements which leads to error.

Versions and main components

• PyMC/PyMC3 Version: 4.0.0b2
• Aesara/Theano Version: 2.3.8
• Python Version: 3.8.12
• Operating system: Windows 10 Pro, Version: 21H1, OS Build: 19043.1566
• How did you install PyMC/PyMC3: conda

Related To: #5544

[ricardoV94]

@Icyshaman Thanks for reporting. Do you want to open a PR to fix this issue?

[Icyshaman]

@Icyshaman Thanks for reporting. Do you want to open a PR to fix this issue?

Sure, I will work on this.

[ricardoV94]

I think we fixed this