postgresql-simple v0.5.0.0 Release Notes

Release Date: 2015-09-19 // over 8 years ago
    • Removed the deprecated BuiltinTypes module.

    • Modified the SQL quasiquoter so that it returns a Query, not an overloaded string, and so that the OverloadedStrings language extension is no longer necessary, thanks to Mike Ledger.

    • Moved away from blaze-builder in favor of bytestring-builder. This shouldn't affect very many people, but does reduce the transitive dependencies.

    • Rewrote the timestamp printers to use the new Prim infrastructure in bytestring-builder. The new printers should be a fair bit faster.

    • Added support for exclusion violations to the ConstraintViolation type in the Errors module, thanks to João Cristóvão.

    • Moved away from the uuid package in favor of the uuid-types package, thanks to Bardur Arantsson. This shouldn't affect anybody, but does reduce the transitive dependencies.

    • Postgresql-simple now explicitly assumes the UTF8 character encoding for communication between the client and server. All database encodings support UTF8 except for Mule Internal Code, the Multilingual Extensions for Emacs. An exception should be raised upon connecting to a database by the backend if the backend cannot accomodate this requirement.

    • Added Eq and Typeable instances for Connection.

    • Added the foldWith, forEachWith, and returningWith families of functions courtesy of Travis Staton.

    • Support for Ranged types, with thanks to Leonid Onokhov for his contributions.

    • The FromField instance for JSON now allows for top-level values that are not objects or arrays, thanks to Sam Rijs.

    • The timestamp parsers have been replaced with those now in Aeson. Janne Hellsten adapted the old parsers from postgresql-simple for inclusion in Aeson; Bryan O'Sullivan rewrote those parsers to be faster, with some tweaks contributed by myself. And now to bring the effort full circle, the result has been brought back to postgresql-simple, with some adaptations.

    • Fixed a bug in the typeinfo system where postgresql's _record type was being reported as a basic type and not an array type. Thanks to Nickolay Kolev for helping to expose this issue.

    • Fixed a bug with the typeInfo operator, thanks to Timmy Tofu. In the case of parsing subfields of arrays and composites, it would fetch the TypeInfo of the array or composite type and not the subtype.