dan2097
Naming of polycyclic ring structures seems to provide unexpected structures
Original report by Anonymous.
The structures provided for known compounds such as 8,9-dihydro-2H,7H-2,9a-diaza-benzo[cd]azulene-1,6-dione and (7-methyl-7,8-dihydro-2H-1,6,9-trioxa-9a-borabenzo[cd]azulen-2-yl)methanamine generate unexpected structures.
Original comment by Daniel Lowe (Bitbucket: dan2097, GitHub: dan2097).
Thanks for the bug report, this issue arises from not all ways of laying out the fused-ring system on an idealised grid being considered (5/7 membered rings are difficult as they can be laid out in multiple ways, and the algorithm prefers to start laying out from a ring with only 1 fused bond, which is not possible in this system). I'll prioritise fixing it as the fact that it gives incorrect numbering is indeed a major bug.
Original comment by Daniel Lowe (Bitbucket: dan2097, GitHub: dan2097).
The particular issue raised here was fixed back on 2017-06-06, but this issue was not closed as more obscure ring systems were found that could still exhibit similar problems, see tripleSubstituedSevenMembered in FusedRingNumbererTest