Re: BUG #17777: An assert failed in nodeWindowAgg.c

Hi,

On 2023-02-10 18:57:10 -0500, Tom Lane wrote:
> I might be more excited about it if there were a visible use-case
> for volatile filter expressions, but I can't see one. The presented
> test case is obviously just a fuzzer, not a useful query.

I don't care about the performance of such a query, but it doesn't seem great
that the defense that we do have, doesn't work. It's not like we don't have a
fallback execution path, it's just that we don't know that we have to use it.

Do you think all other uses of contain_volatile_functions() (and it looks like
contain_mutable_functions()) are fine with not detecting volatility in
subplans?

I don't think it's common, but I don't think it's crazy to have a volatile
function somewhere within an aggregate. pg_current_wal_lsn(),
clock_timestamp(), pg_relation_size() or such.

I think we could just add a !contain_subplans() to the code deciding whether
it's safe to use the movable window optimization?

Greetings,

Andres Freund