Cannot overwrite the __str__ method in jitclass

Hello,

I am not able to overwrite the string representation of a jitclass. I do not know if this is a wanted behaviour though.

from numba.experimental import jitclass
from numba import float64

spec = [("alpha", float64)]


@jitclass(spec)
class Foo:
    def __init__(self, alpha):
        self.alpha = alpha
    
    def __str__(self):
        return "Class Foo"


foo = Foo(1.)

print(str(foo))

Produces:

<numba.experimental.jitclass.boxing.Foo object at 0xS0M3wh3r3>

Expected:

Class Foo

It seems that the __str__() method is skipped entirely, in fact the following does not produce any ZeroDivisionError.

...
    def __str__(self):
        return 0 ** (-2)

Tested on:

Platform: Linux-5.15.0-52-generic-x86_64-with-glibc2.35
Python version: 3.9.7 (default, Sep 20 2022, 16:21:01) 
[GCC 11.2.0]

Numba version: 0.55.2

[stuartarchibald]

Thank you @stuartarchibald! After updating to 0.56.3, the __str__ method works!

My bad, I did not notice it from the Release notes

No problem, glad it is working now.

Next question: what about __repr__? Any possibility for it to be implemented in a future release? It’d be really nice to have it for usage within Jupyter Notebook.

I suggest opening a feature request if it is important to you. I think it’s technically possible to support it.

[stuartarchibald]

Done!

Ref. #8533

Great, many thanks. I’ll close this issue as resolved.