Warren Weckesser

@lzha97, there must be a blank line at the beginning and end of each block of interactive shell prompts, e.g. ``` A real world example: Assume a company has 10000...

It looks like this doesn't catch all the cases. The following is run with this pull request: ``` In [2]: import scipy In [3]: scipy.__version__ Out[3]: '1.10.0.dev0+1308.ca45421' In [4]: from...

I was running through some more examples, and ran into this: ``` In [71]: csr_matrix([[5, 5, 3], [4, 9, 10], [3, 4, 9]]).argmin() # Should be 2 Out[71]: 6 ```...

I can confirm that this crashes Python with the main development branch, too. Here's a simplified reproducer: ``` import numpy as np from scipy.stats import ncf x = np.linspace(0, 1,...

This is a consequence of how NumPy stores strings (Unicode string and `bytes`). In a NumPy array with dtype, say, `U6`, there is storage allocated for 6 Unicode characters for...

Thanks for digging into this, @rgommers. I'll try installing the Intel oneAPI toolkit this weekend, and if that goes well, I'll look into some of these failures.

I ran into the Pythran issue too. To get the build to succeed, I installed Pythran from the git branch `feature/0.12.0`. I get a different mix of failures than those...

> ... are all other modules failure-free for you? No, there are other failures. I see failures in `odr`, `optimize` and `special`, and multiple Python worker crashes in `linalg` and...

With the main branch (commit 0fa0a40b3f), I now get 7 failures in `special`: short test summary info ``` ======================================== short test summary info ========================================= FAILED scipy/special/tests/test_exponential_integrals.py::TestExp1::test_branch_cut - AssertionError: a... FAILED...

Thanks @eli-schwartz, I created a pull request: https://github.com/scipy/scipy/pull/17109.