Bug report
Bug description:
I just encountered the situation where I used random.sample but both the population and counts arguments were empty (my algorithm had nothing left to choose from). So, basically this situation:
>>> random.sample([], 1, counts=[]) Traceback (most recent call last): File "<python-input-1>", line 1, in <module> random.sample([], 1, counts=[]) ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^ File "/path/to/lib/python3.14/random.py", line 424, in sample total = cum_counts.pop() IndexError: pop from empty list
Instead of the IndexError, I expected a ValueError, similar to the following situations:
>>> random.sample([], 1) Traceback (most recent call last): File "<python-input-2>", line 1, in <module> random.sample([], 1) ~~~~~~~~~~~~~^^^^^^^ File "/path/to/lib/python3.14/random.py", line 434, in sample raise ValueError("Sample larger than population or is negative") ValueError: Sample larger than population or is negative >>> >>> random.sample([1], 2, counts=[1]) Traceback (most recent call last): File "<python-input-3>", line 1, in <module> random.sample([1], 2, counts=[1]) ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^ File "/path/to/lib/python3.14/random.py", line 429, in sample selections = self.sample(range(total), k=k) File "/path/to/lib/python3.14/random.py", line 434, in sample raise ValueError("Sample larger than population or is negative") ValueError: Sample larger than population or is negative
The docs mention that
If the sample size is larger than the population size, a ValueError is raised.
In addition, I would expect the following to work:
>>> random.sample([], 0, counts=[]) Traceback (most recent call last): File "<python-input-4>", line 1, in <module> random.sample([], 0, counts=[]) ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^ File "/path/to/lib/python3.14/random.py", line 424, in sample total = cum_counts.pop() IndexError: pop from empty list
similar to how it works when counts is not specified:
>>> random.sample([], 0) []
Not sure though what CPython's backwards-compatibility policy has to say here, since changing the exception type – or, in the second case, removing the exception altogether – might actually break someone's code...
Tested with:
Python 3.14.0a5 (main, Feb 12 2025, 14:51:40) [Clang 19.1.6 ] on linux
cpython-3.14.0a5-linux-x86_64-gnu
CPython versions tested on:
3.14
Operating systems tested on:
Linux