Sergei Petrunia

MySQL 8.0 has "anti-semi-join" optimization: https://mysqlserverteam.com/antijoin-in-mysql-8/ It is available since 8.0.17 (the same version that https://github.com/facebook/mysql-5.6/tree/fb-mysql-8.0.17 is based on). Anti-semi-join handles WHERE ... NOT IN /NOT EXISTS queries. The EXPLAINs...

Yes, if the subquery is in the select list, then batching is not possible. Possible ways to solve this: - Batching for subqueries in the select list - Merge the...

https://github.com/facebook/mysql-5.6/pull/1165#pullrequestreview-769443923 > The following doesn't look right: > ... I can't reproduce this behavior... This is what I'm getting: ``` mysql> begin; Query OK, 0 rows affected (0.00 sec) mysql>...

> Another example: > > ``` > mysql> create table t6 (pk int not null auto_increment, a int, b int, c int, primary key (pk), unique key (a), key (b),...

Double checking if I am using the right tree... it is https://github.com/spetrunia/mysql-5.6/commits/fb-mysql-8.0.20-range-locking-may ``` commit 9f29d708d47db4aaf4e015697dd088d2ee408b44 (HEAD -> fb-mysql-8.0.20-range-locking-may, origin/fb-mysql-8.0.20-range-locking-may) Author: Sergei Petrunia Date: Mon Sep 13 20:23:12 2021 +0300 Range...

lock escalation is triggered by lock tree memory exceeding the limit. The current lock memory usage should be shown in `rocksdb_current_lock_memory`, and `rocksdb_max_lock_memory` is where one sets the limit.

This happens only when the used index has columns with certain datatypes. If one tries to run the same example with e.g. integer: ```c++ create table t2 ( pk int...

One other question: What if Item_const::fix_fields() always did the following, regardless of its location: * call `value_item->set_name(...)` in the same way it does for itself in `Item_const::Item_const`. * substitute itself...

I don't understand how the last variant is supposed to work? I see this check ``` if (thd->last_sql_command == SQLCOM_SELECT && ``` what is it supposed to achieve?

As far as I understand, the issue raised in https://jira.mariadb.org/browse/MDEV-33971?focusedCommentId=285058&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-285058 is still not addressed. Double-checking this. I take ``` commit 05e9b4a6955bcbedc237e46fafdccb2b4419d1a6 (HEAD -> 10.6-mdev-33971, origin/10.6-mdev-33971) Author: Dave Gosselin Date: Fri...