Adjust Python bindings for SWIG >= 4.4.

[aryler]

Function signature changed in SWIG 4.4. Details:
numpy/numpy@1e424da

Could maybe use NULL instead of nullptr; they should be the same with SWIG < 4.4.
Minor detail. Not sure what's better.

[Krakonos]

Hi!

Let's say the API change is a bit unfortunate. Based on the commit mentioned, the 0 is correct. However, I believe the previous version should use NULL instead of nullptr, since nullptr is a C23 feature ( https://en.cppreference.com/w/c/language/nullptr.html ). It might be mistaked for nullptr in C++ that was introduced much earlier, but the code in question is C.

[aryler]

Hi, thanks for looking into this.

Here, you can see similar, using NULL. They haven't removed the Python 2 guard (yet?).
coin3d/pivy@b8e55e2

[Krakonos]

Hi! Yes, that's correct. But the PR as submitted uses nullptr (which was inherited from the original code, but I think it's worth fixing as well).

[Krakonos]

The change is correct. The only reservation I have is to fix the nullptr -> NULL, but that was not the scope of this PR.

[johnbeard]

Based on the change that introduced this in Swig 4.4, I think it's expected to return -1 on failure rather than 0: swig/swig@6c3bc2d

Because the value will be used in this if:

  if (SWIG_mod_exec(m) != 0)
  {
    Py_DECREF(m);
    return;
  }