mirror of
https://github.com/processone/ejabberd.git
synced 2024-06-08 21:43:07 +02:00
22d49ad2fe
- Conversion of UNIQUE KEY constraints to PRIMARY KEY - in pg a PRIMARY KEY is a UNIQUE KEY where all columns are non-null, so this patch makes it more explicit. - Removal of useless indexes - a btree index of (A,B,C) implies fast indexing of (A,B) and (A) but directly specifying these indexes causes much IO overhead on record modification - Allow more than 2b items in spool/pubsub by switching SERIAL to BIGSERIAL - Add some TODOs where it looks like columns should be marked NOT NULL but are not actually. As I'm not particularly familiar with the ejabberd I don't know if this should change or not. |
||
---|---|---|
.. | ||
lite.new.sql | ||
lite.sql | ||
mssql.sql | ||
mysql.new.sql | ||
mysql.old-to-new.sql | ||
mysql.sql | ||
pg.new.sql | ||
pg.sql |