fix: boost - clouds on the horizon

[sbillinge]

Problem

pyhthon boost is going out of fashion with the speedup changes in the kernel taking place in 3.13, 3.14 .... and will get more and more difficult to builds and likely stop working altogther shortly.

Proposed solution

migrate everything in the diffpy codebase (including libdiffpy/pyobjcryst to the new approaches, either pybind11 as a kind of drop-in replacement or, even better, nanobind

@sbillinge is about to try and recruit a new cohort of developer wizards to replace our existing hard working cohort who will be going on to greater things. This could be a good task for them.

We have suspended trying to release our c++ codes beyond 3.13. Though it is technically possible, it may not be worth it because a single dependency that won't build kind of breaks the chain.

@vincefn any thoughts on this?

[vincefn]

Hi @sbillinge - I guess this was coming, boost is a rather old binding implementation. Nothing urgent, we can live for a year or so without py3.14 support but this'll need to be addressed.

As for the new backend, I have no apriori - I see however in the nanobind documentation:

"pybind11 must deal with all of C++ to bind legacy codebases, while nanobind targets a smaller C++ subset. The codebase has to adapt to the binding tool and not the other way around, which allows nanobind to be simpler and faster"

So we'd have to check that nanobind can handle all the ObjCryst++ API.

Still would be very time-consuming (multi-platform build & checks), so we have to see if boost-python is really stuck.

Then again, maybe an AI bot would make the conversion trivial.

[sbillinge]

no rush on this, but it would be good to work on this this year. We can pin everything that has c-code or inherits pyobjcryst to 3.13 in the meantime