Nathan Goldbaum
Nathan Goldbaum
I made #339
I wonder what happens if you plot the result of your benchmark as a function of thread count. See [e.g. this NumPy issue](https://github.com/numpy/numpy/issues/27786) which reported a similar scaling issue and...
Sorry for duplicating the open issue. Now that virtualenv has done the release this is unblocked.
I tried running the `tox` test suite on a free-threaded Python, but that crashes because `devpi` depends on `argon2-cffi-bindings`, which itself depends on `cffi`. I think we'll have to wait...
Thanks so much for finishing this off @gaborbernat.
It would be great if Cython provided this. Doing it in standard C is a little annoying, depending on how old your compiler is.
Sayed has some ideas along these lines in his C++ refactor: https://github.com/numpy/numpy/pull/19753. I'm not sure how much appetite there is to add more things to libnpymath. That said, I see...
@crusaderky can you give a little extra context for how versioned_hdf5 uses array hashes?
Thanks for thinking about this a little more carefully than me. I think having some sort of API for producing hashable data makes sense, along with a more general scheme...
@MaanasArora, @seberg and I chatted a little bit privately about this and we now think a better way to go with this is having a well-defined, versioned schema that can...