Oyibo

Unfortunately, the implementation of numpy.fromiter seems to be problematic. 1) The output array shape and type must be determined by a function argument "dtype" not by an argument type. ```...

@gmarkall no problem, I will open another issue for advanced use of numpy.dtype as feature request https://github.com/numba/numba/issues/9527 What about the issue in np.asarray where a typed list is not identified...

@gmarkall thank you. The implementation of numpy.asarray converts a typed list of scalars into a numpy array. The function fails converting a typed list of 1D-numpy arrays into a 2D-numpy...

@gmarkall thank you. If asarray would be able to convert a typed list of 1D-arrays into a 2D-array that could theoretically make room to implement np.fromiter. Although I'm not sure...

@dschmitz89 thank you for the information. At the moment the setup.py file of numba-scipy has a pinned dependency on scipy= 2.0, Numba >= 0.60, SciPy >= 1.14, or Python >=...

@ilayn, thank you for the clarification. As I understand it, the typical approach still involves wrapping the scipy.special functions using C/C++ or Cython, compiling them into a shared library, and...

@rgommers Thank you for your proposal. In many cases, it's preferable to use ahead-of-time compiled or at least cached Numba functions. However, caching compiled Numba functions isn't possible in your...

A potential workaround (other than C-wrappers) would be to retrieve the physical memory location of the SciPy function and use it within the Numba jitted function. This approach, however, could...

> I'm thinking if there's a special opt-in for shared libraries like load_library_permanently to enable caching, maybe that should exist for Cython APIs as well? That will work within the...

@DWesl Thank you for your proposals. There are multiple ways to interact with C/C++ libraries from Numba, each with varying degrees of ease and cachability. I'm not an expert but...