Merged PRs

dolt

• 6000: Evaluate check constraints during merge
  Adds support for evaluating check constraint expressions on a table during a merge.
  Fixes #3462
• 5998: Add docstring to ConnectionQueryist.

go-mysql-server

• 1790: Reverting @@system_time_zone support for returning system timezone
  I'm not sure yet why this change causes Dolt's MySQL Connector J integ test to fail, but after testing commits yesterday and testing this fix this morning, this is definitely causing some problem. In CI, but not locally, the MySQL Connector J integ test is failing with a generic/empty SQLException each time, immediately after executing the first query. Dolt logs look fine, but there's obviously something happening that the MySQL Connector J can't deal with. I have a few more repro ideas that I'll follow up with, but wanted to get the GMS -> Dolt pipeline cleared up first.
  Until then, this PR reverts @@system_time_zone from returning the actual timezone in use by the runtime and OS.
  This change was able to get the Dolt MySQL Connector integ suite passing again: https://github.com/dolthub/dolt/actions/runs/5059419161/jobs/9081665231
• 1789: foreign key index match should be on prefix not subset
  Turns out the logic for using an existing secondary index for foreign keys was matching on a subset instead of a prefix.
  For example:

  fks> create table parent (fk1 int, fk2 int, primary key (fk1, fk2));
  fks> create table child (id int unique, fk1 int, fk2, primary key (fk2, fk1, id), foreign key (fk1, fk2) references parent (fk1, fk2));
  

  This should produce a child table that creates a new secondary index over the columns (fk1, fk2).
  But, we don't do that and instead reference some other index (not sure what), causing errors when calling dolt constraints verify --all
  This PR ensures that we only pick an existing secondary index as the underlying index for a foreign key if that index is a prefix for the foreign key.
• 1785: Copy MySQL information_schema.st_spatial_reference_systems table
  This PR essentially copies the information_schema.st_spatial_reference_systems table from MySQL.
  Small refactor to move constants and help function to types package from functions/spatial package.
  Additionally, changes show create table statement to print SRIDs inside a MySQL special comment.
  Example:

  tmp> show create table t2;
  +-------+------------------------------------------------------------------+
  | Table | Create Table                                                     |
  +-------+------------------------------------------------------------------+
  | t2    | CREATE TABLE `t2` (                                              |
  |       |   `p` point /*!80003 SRID 2000 */,                               |
  |       |   `l` linestring /*!80003 SRID 2001 */                           |
  |       | ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_bin |
  +-------+------------------------------------------------------------------+
  1 row in set (0.00 sec)
  

  Partially Addresses this: #5973
• 1781: Small aggregation perf improvements
  • We used to make a new error object for every cache.Get() miss. Use shared object instead.
  • Add reuse pool for collation weight buffers rather than making thousands of [4]byte.
    before:
  after:

vitess

• 238: Fix for charset introducers in default values
  Fixes #5970 by adding an additional default expression rule that handles charset introducers.
• 237: Small fix for DEFAULT CHARSET when creating tables
  Super small fix for #5749. We were referencing the wrong variable.
• 233: support parsing ALTER EVENT statements
  Supports parsing ALTER EVENT statements

Closed Issues

• 3462: dolt merge silently violates check constraints
• 1771: utf8mb3_tolower_ci not implemented

Latency