/ 
DBIO-MySQL-0.900004
/ DBIO::MySQL::Storage::MariaDB

NAME

DBIO::MySQL::Storage::MariaDB - MariaDB-specific storage for DBIO

VERSION

version 0.900004

SYNOPSIS

package MyApp::Schema;
use base 'DBIO::Schema';
__PACKAGE__->load_components('MySQL::MariaDB');

my $schema = MyApp::Schema->connect($dsn, $user, $pass);

DESCRIPTION

MariaDB-specific storage backend for DBIO: the DBIO::MySQL::Storage subclass registered for a dbi:MariaDB: DSN (DBD::MariaDB). It overrides no behaviour of its own. It exists as the registration target for the MariaDB DBI driver, and as the class the future_io transport is resolved from by convention (DBIO::MySQL::Storage::MariaDB::Async).

The DBD::MariaDB attribute names -- mariadb_insertid for the last insert id, mariadb_auto_reconnect for the auto-reconnect that is switched off on every connect -- are not chosen here. DBIO::MySQL::Storage reads them off the DBD that owns the handle ("_dbd_attribute" in DBIO::MySQL::Storage). load_components('MySQL') over a dbi:MariaDB: DSN never reaches this class, so a choice made here would miss it. This class used to override _dbh_last_insert_id and _run_connection_actions for them. The second override skipped the parent's _run_connection_actions on purpose, and with it anything the parent added there. Both are gone.

This class is auto-registered for the MariaDB DBI driver and is activated when "connection" in DBIO::MySQL::MariaDB is called.

This class is the DBD, not the engine

Everything above is a DBD::MariaDB fact, and that is the whole of what this subclass is for. It is not a statement about the server at the other end: DBD::MariaDB connects to MySQL servers just as happily, and a MariaDB server reached through a dbi:mysql: DSN lands in the parent class instead.

So no SQL dialect decision is taken here at class level, and none is left: which spelling of an upsert's attempted values the server understands, whether it has RETURNING, which locking-read grammar it has, and which spelling of the replica-status statement it answers are all properties of the engine, and all are decided in DBIO::MySQL::Storage from "_is_mariadb" in DBIO::MySQL::Storage plus the server version -- correct for all four DBD/engine combinations, where a class-level answer here is right for two of them and wrong for the other two.

Two of those are not decided here, and each would be a latent syntax error of that exact shape:

  • The locking-read grammar is not set by this class's sql_maker_class (pointing at DBIO::MySQL::SQLMaker::MariaDB purely to spell for => 'share' as LOCK IN SHARE MODE). A MariaDB server reached through a dbi:mysql: DSN never gets here at all and would be handed FOR SHARE, which MariaDB has in no version; a MySQL 8 reached through DBD::MariaDB -- how this dist's live suite reaches MySQL -- does get here and would be handed LOCK IN SHARE MODE OF `t` NOWAIT, which MySQL rejects, because the legacy spelling is precisely the one that carries neither OF nor a modifier. See "_lock_dialect" in DBIO::MySQL::Storage.

  • The replica-status statement is not overridden here to try SHOW REPLICA STATUS first. The rename is a server fact on both products and they made it at different releases -- and MySQL renamed the output columns with it, so against a MySQL 8.0.22+ replica through DBD::MariaDB the statement would succeed and is_replicating would then read a Slave_IO_Running that was no longer there. Meanwhile the parent's plain SHOW SLAVE STATUS is a syntax error on MySQL 8.4, which removed it. See "_replication_status_statement" in DBIO::MySQL::Storage.

SEE ALSO

AUTHOR

DBIO & DBIx::Class Authors

COPYRIGHT AND LICENSE

Copyright (C) 2026 DBIO Authors Portions Copyright (C) 2005-2025 DBIx::Class Authors Based on DBIx::Class, heavily modified.

This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.