"Etsuro Fujita" <fujita.etsuro@lab.ntt.co.jp> writes:
> [ pathkey_and_uniqueindx_v7_20131203.patch ]
I started to look at this patch. I don't understand the reason for the
foreach loop in index_pathkeys_are_extensible (and the complete lack of
comments in the patch isn't helping). Isn't it sufficient to check that
the index is unique/immediate/allnotnull and its ordering is a prefix
of query_pathkeys? If not, what's the rationale for the specific tests
being made on the pathkeys --- this code doesn't make much sense to me.
regards, tom lane