Re: Expression errors with "FOR UPDATE" and postgres_fdw with partition wise join enabled.

On Fri, Aug 3, 2018 at 10:07 AM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Robert Haas <robertmhaas(at)gmail(dot)com> writes:
>> On Thu, Aug 2, 2018 at 4:25 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>>> Now, that's a bit of a problem for postgres_fdw, because it seems to
>>> insist on injecting WRVs even when the query text does not require any.
>>> Why is that, and can't we get rid of it?
>
>> I don't quite know what you mean here -- postgres_fdw does use
>> whole-row vars for EPQ handling, which may be what you're thinking
>> about.
>
> As far as I can see from the example that started this thread,
> postgres_fdw injects WRVs into a PWJ whether or not the query involves
> FOR UPDATE; that's why this bug is reproducible in a query without FOR
> UPDATE. But we shouldn't need any EPQ support in that case.

I might be missing something, but the original reproducer involves a
FOR UPDATE clause, and the expected regression test output in
postgres_fdw.out appears to show a whole-row var in the foreign scan's
target list only in those examples where a locking clause is present.

>> Honestly, I'm pretty impressed that we have added not one but two
>> members to the RelOptKind enum without as little collateral damage as
>> there has been.
>
> Color me a bit more skeptical about the bug density in that, given
> that enable_partitionwise_join is off by default; that means you're
> not getting a lot of testing.

You're hard to please.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company