Disallow redundant bonds in BondList

[Linford24]

No description provided.

[Linford24]

A check for same atom binding to itself was added to both the BondList constructor and add_bond method.

[padix-key]

Addresses first part of #812

[padix-key]

Thanks for the fix. However, I see a problem with the performance: Iterating over bonds directly happens in pure Python, which can be quite time consuming. Hence I'd rather suggest to use a memoryviewinstead which cython compiles into C code.

Are you familiar with these Cython details? Otherwise I also can make the amendments here.

[Linford24]

Yes please. This PR only addresses the first part. The second part regarding the chem_comp bond will be in a separate PR as you suggested and I will be push soon

[Linford24]

I am not familiar with the Cython details but I would be glad if you could brief me on them so I implement the suggestion myself and get to learn it as well

[Linford24]

Meanwhile, I am reading about the change you suggested. Would get back with the revised amendments

[Linford24]

Added iteration over bonds with memoryview instead of pure Python

[padix-key]

Sorry for the late response, I was sick for a few days. I think these two tutorial pages give a sufficient overview of Cython's memory views:

https://cython.readthedocs.io/en/latest/src/userguide/numpy_tutorial.html
https://cython.readthedocs.io/en/stable/src/userguide/memoryviews.html

If you look at bonds.pyx you will often encounter lines, like these:

cdef uint32[:,:] all_bonds_v = self._bonds

It tells Cython to convert self._bonds into a typed memoryview of uint32 values, which can then be efficiently accessed in C-code, without expensive calls to the Python API. You can access the dimensions of the memoryview using .shape and use integer indices to access specific elements. Note that fancy indexing is not possible for these constructs

[Linford24]

I'm much glad you're fit again. I'm going through the tutorials