Ilhan Polat
Ilhan Polat
Indeed things are quite straightforward but as with other fortran swaps, the really difficult part is to follow the memory pointeers such that we don't break too much code. However...
Oh and if you have such problems lying around, I'd really appreciate if you also test them, with and without consistent constraints.
I finished SLSQP conversion but there are so many tiny helper functions piled up on top of the core function, I can't figure out how to get things out of...
> Because you've been diving deep into this code it'd be helpful if you could share any insights you have on what xk might represent at the end of each...
Thanks @antoinecollet5 indeed we have been discussing this with him over Fortran discourse too. But I already finished this one locally. The main issue is how to integrate to our...
Superseded by #22524
One particular con about nanobind is the cmake dependency which we don't need to discuss again why we don't prefer it.
I can't wrap (excuse the pun) my head quite yet what is going on there but I'm just trying to supply sufficient information to meson and crossing fingers.
That is to wrap C/C++ code wrap in a more lightweight and more performant fashion. Cython is indeed adding quite some beef into the binaries and I'm not sure about...
If I take the benchmarks at the face value, build times and binary sizes are really appealing given how much cython wrapped code we have including against pybind11. I can...