Robert Haas <robertmhaas@gmail.com> writes:
> I have cleaned up and simplified this patch. Attached is the version
> I intend to commit. Changes:
Minor suggestions:
I think the names like "rel_parents" would read better as
"rel_numparents" etc. As-is, the reader could be forgiven for expecting
that this will be a list of parent relation OIDs or some such.
The new loop added within find_all_inheritors could really do with an
addition to the comments, along the line of "If a child is already
seen, increment the corresponding numparents count".
I don't trust the proposed "order by attrelid" business in the
regression test --- once in a blue moon, that will fail because the
OID counter wrapped around mid-test, and we'll get an unreproducible
bug report. I'd suggest order by attrelid::regclass::text.
Looks sane otherwise.
regards, tom lane