[SMARTS] Ambiguious behavior of tetrahedral centers placed adjacent to ring closures

[b-mahjour]

I have a pair of SMARTS that I suspect are equivalent but are not recognized as such by RDKit.

On the left, a counter-clockwise carbon is placed before a ring closure, and we get the corresponding projection with the first atom being [CH] since it's attached to the ring closure. On the right, the ring closure is behind a clockwise carbon, and so it is bonded to [CH]. Here is where I am confused as it seems that the ring closure should be ignored and produce the projection I've drawn, which is equal to the conformation on the left.

from rdkit import Chem

patt_a = Chem.MolFromSmarts("O[C@]1C[CH]1N")
patt_b = Chem.MolFromSmarts("N[CH]1([C@@](O)C1)")

s_mol = Chem.MolFromSmiles("O[C@H]1CC1N")

print(s_mol.HasSubstructMatch(patt_a, useChirality=True))
print(s_mol.HasSubstructMatch(patt_b, useChirality=True))

Results in

False
True

Despite this, they fail to perform equally when matching a molecule.

ChemDraw seems to agree

I would appreciate it if anyone would please let me know if I am missing something here or point me toward a resource where the behavior of chiral tags adjacent to ring closures are described.