Skip to content

Conversation

danolivo
Copy link
Collaborator

@danolivo danolivo commented Apr 7, 2023

No description provided.

danolivo and others added 4 commits April 7, 2023 14:56
It mostly caused by desire of reducing number of failures 001_pgbench.pl test
on WINDOWS OSes (it is related to speed of file descriptor allocations in
the test, where we CREATE/DROP extensions competitively by several threads.

Also, the aqo_CVE-2020-14350 test is corrected.
different libraries.

To avoid such a problem in future, refactor AQO interfaces: declare all
hooks as static, reduce number of exporting functions and introduce
concept of *_init() function for a module that needs some actions in the
PG_init() routine.

Reviewed by: @Anisimov-ds

P.S.: being cherry-picked from stable14 the commit changed drastically because
of huge differences between the PG cores...
…n of AQO prediction hooks. It isn't a strict rule, but we should know about that.
Also, some arrangement for stable14 added by a.lepikhov
@danolivo danolivo added this to the AQO 1.6 milestone Apr 7, 2023
@danolivo danolivo self-assigned this Apr 7, 2023
@danolivo danolivo merged commit 5b215f2 into stable13 Apr 8, 2023
@danolivo danolivo deleted the catch-up-stable14 branch April 8, 2023 01:36
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant