Be notified of new releases
Create your free GitHub account today to subscribe to this repository for new releases and build software alongside 28 million developers.Sign up
- Add support for hypothetical partitioning, for pg10+ (Hosoya Yuzuko,
- Use a dedicated MemoryContext to store hypothetical objects
- Call previous explain_get_index_name_hook if it was setup
- add hypopg_reset_index() SQL function
- Add support for INCLUDE on hypothetical indexes (pg11+)
- Add support for parallel hypothetical index scan (pg11+)
- Fix support for pg11, thanks to Christoph Berg for the report
- Fix potentially uninitialized variables, thanks to Jeremy Finzel for the
- Support hypothetical indexes on materialized view, thanks to Andrew Kane
for the report.
- add support for PostgreSQL 11
- add support for hypothetical indexes on expression
- add a hypopg_get_indexdef() function to get definition of a stored
- don't allow hypothetical unique or multi-column index if the AM doesn't
- disallow hypothetical indexes on system columns (except OID)
- fix indexes using DESC clause and default NULLS ordering, thanks to
Andrew Kane for the report and test case.
- fix PostgreSQL 9.6+ support, thanks to Rob Stolarz for the report
- add support for PostgreSQL 10
- add support for PostgreSQL 9.6, thanks to Konstantin Mosolov for fixing some
- add support from new bloom access method (9.6+)
- fix issue with hypothetical indexes on expression (thanks to Konstantin
- fix possible crash in hypothetical index size estimation
- fix compatibility with pg 9.4-
- add support for BRIN indexes (for pg 9.5+ obviously).
- handle index on predicate.
- handle index storage parameters for supported index methods. For now, this means :
- fillfactor for btree indexes
- pages_per_range for brin indexes
Bug fixes or improvements:
- check that number of column is suitable for a real index
- free memory when hypothetical index creation fails
- for btree indexes, check that the estimated average row size is small enough to allow a real index creation.
- safer handling of locks.
- remove the simplified "hypopg_add_index()" function
- fix a bug when a regular query could fail after a hypothetical index have been created, and tested with explain.
- hypopg_create_index() and hypopg_add_index() now returns the oid and index names.
- add hypopg.enabled GUC. It allows disabling HypoPG globally or in a single backend. Thanks to Ronan Dunklau for the patch.