CanonicalRankAtomsInFragment example in the documentation is not reproducible

[chmnk]

Describe the bug
The documentation's example of CanonicalRankAtomsInFragment produces a different output compared to the documentation.

To Reproduce
In the documentation:

>>>mol = MolFromSmiles('C1NCN1.C1NCN1')
>>>list(CanonicalRankAtomsInFragment(mol, atomsToUse=range(0,4), breakTies=False))
[0,1,0,1,-1,-1,-1,-1]
>>>list(CanonicalRankAtomsInFragment(mol, atomsToUse=range(4,8), breakTies=False))
[-1,-1,-1,-1,0,1,0,1]

My RDKit's behaviour:

>>>mol = Chem.MolFromSmiles('C1NCN1.C1NCN1')
>>>print(list(Chem.CanonicalRankAtomsInFragment(mol, atomsToUse=range(0,4), breakTies=False)))
[4, 6, 4, 6, -1, -1, -1, -1]
>>>print(list(Chem.CanonicalRankAtomsInFragment(mol, atomsToUse=range(4,8), breakTies=False)))
[-1, -1, -1, -1, 4, 6, 4, 6]
>>>print(list(Chem.CanonicalRankAtomsInFragment(mol, atomsToUse=range(0,4), breakTies=True)))
[4, 6, 5, 7, -1, -1, -1, -1]
>>>print(list(Chem.CanonicalRankAtomsInFragment(mol, atomsToUse=range(4,8), breakTies=True)))
[-1, -1, -1, -1, 4, 6, 5, 7]

Expected behaviour
Actually I find the example not very intuitive, cause in both cases breakTies is set to True. So I don't know what behaviour is expected.

Configuration (please complete the following information):

• RDKit version: both '2023.03.1pre' and '2022.09.3'.
• OS: CentOS Linux 7
• Python version: 3.10 and 3.8
• Are you using conda? Yes
• If you are using conda, which channel did you install the rdkit from? '2022.09.3' from conda-forge
• If you are not using conda: how did you install the RDKit? '2023.03.1pre' from sources inside of a conda environment

[greglandrum]

Confirmed. Thanks for the bug report

[chmnk]

What behaviour is expected by the way? The same as in CanonicalRankAtoms, but with -1 on unlisted atoms, or something else?

[greglandrum]

I think the behavior of the code is fine since the relative ranking of the atoms is correct. The documentation is incorrect and should be fixed.