On Tue, May 10, 2022 at 04:09:47PM +0900, Michael Paquier wrote:
> Well, what about the attached then? While looking at all the headers
> in the tree, I have noticed that __pg_log_level is not marked, but
> we'd better paint a PGDLLIMPORT also for it? This is getting used by
> pgbench for some unlikely() business, as one example.
After an extra look, PGDLLIMPORT missing from __pg_log_level looks
like an imbroglio between 8ec5694, that has added the marking, and
9a374b77 that has removed it the same day. All that has been fixed in
5edeb57.
--
Michael