1.4.2

• fix reconcile-sequence-for-column! logic to account for is_called behavior.
• fix connection pool error message log to include the pool's name

1.4.1

• no changes, accidental release

1.4.0

• move to HikariCP 5.x, minor setHealthCheckRegistry instantiation change to accomodate

1.3.0

• update java.jdbc to 0.7.11
• update migratus to 1.3.5. This version of migratus has breaking changes.
• migrations with multiple statements that lack a separator will now fail
• change the wrap-with-delayed-init function to handle migrations from multiple directories

1.2.5

• alter the uncompleted-migrations function to not have any side-effects. Prior to this change, the routine would create the migraiton table if it didn't exist.

1.2.4

• utility methods for transforming data to PGObject jsonb format and reversing the transformation

1.2.3

• wait until migrations have reached a safe point before closing the pool. If the pool is closed during a migration step, the migration is marked as reserved and won't be able to continue.

1.2.2

• update to migratus 1.0.8
• change migration function to use interruptibility built into migratus.

1.2.1

• alter migration function to observe interruptibility of the thread so that migrations can be interrupted
• cleanup connection used in uncompleted-migrations function to prevent leak.

1.2.0

• alter pool initialization behavior to retry transients, but fail on everything else.
• alter pool getConnection to throw a RuntimeException when trying to use the pool with an initialization failure
• alter pool behavior to run migrations when replication-mode is left out of the migration options
• multiple test flakeyness issues resolved

1.1.1

• fix issue with pg-logical wrappers and multiple statements issued by migratus

1.1.0

• Update libraries for Java 9 compatibility
• update migratus to version 1.0.3
• update postgresql driver to 42.2.0
• update java.jdbc to 0.7.5

1.0.3

• update HikariCP to version 2.7.4
• update postgresql driver to 42.1.4

1.0.2

• fix deprecation warning on 'initializationFailFast' in Hikari configuration

1.0.1

• fix an issue where exceptions during migration for a replica were not caught

1.0.0

• update HikariCP to version 2.6.1
• update kitchensink to 0.3.0
• update cheshire to 5.7.1
• update dev dependency slf4j related items to 1.7.22 to sync with Hikari

0.6.2

• Fix create-db! when the owner of the new database is a different user than the one creating the database.
• Add has-role?.

0.6.1

• Improved Japanese translation strings.

0.6.0

• Add db-exists?, create-db!, and drop-db!.
• Add user-exists?, create-user!, and drop-user!.

0.5.0

• Add lifetime protocol to pool interface to allow shutdown routines to wait for initialization to complete.

0.4.15

• Fix consolidate-provider-status to correctly report the status.

0.4.14

• Fix an issue where the Hikari config option setInitializationFailFast was set after the connection was made, which meant that the option wasn't applied to the connection.

0.4.13

• Change the behavior of reconcile-sequence-for-column! so that it will never set a sequence to a lower value than it currently has.
• Update the migratus dependency to 0.8.30 (no breaking changes).

0.4.12

• Remove the NOWAIT argument when locking a table for reconcile-sequence-for-column!. This is unlikely to do anything but cause problems in most circumstances.

0.4.11

• Fixed a bug with spec->migration-db-spec that would allow nil :password keys to appear in the output. The same function now also properly separates :user from :migration-password and :migration-user from :password, so that the two sets of credentials won't be erroneously combined.

0.4.10

• Added the ability to pass in a timeout to combined-replication-status (defaulting to four seconds) which will cause the replication status to be unknown if the replication status cannot be retrieved in the span of the timeout. This fixed a bug with replication-status which caused trapperkeeper-status checks to timeout when no database connection was available because the replication status check was not timing out.

0.4.9

• Fix a bug with reconcile-sequence-for-column! that prevented it from working when the given column was empty.

0.4.8

• Block startup of :replication-mode replicas waiting on migrations. If the application is configured as a replica wait on migrations before initializing the database (useful for waiting on pglogical replication which will replicate the migrations).
• Added Add reconcile-sequence-for-column! that will reset a sequence to the maximum value in the column, or else 0.

0.4.7

• Add support for running migrations when creating a connection pool (separate from the init-fn).
• Add functionality for retrieving the status of pglogical replication.
• Add drop-public-functions! utility function.

0.4.6

• Allow the connection health check timeout to be set via options->hikari-config using the :connection-check-timeout entry.
• Select the :connection-check-timeout entry in select-user-configurable-hikari-options.

0.4.5

• Change exception behavior of update-pglogical-replication-set. If the db user doesn't have rights to update pglogical, catches the exception and returns false. Other exceptions are passed through.

0.4.4

• Add select-user-configurable-hikari-options to provide a consistent way of preserving only the hikari specific options from a given map of options.

0.4.3

• Specifying the timeout in wrap-with-delayed-init does not allow the connectivity timeout to be configured independently. Added add-connectivity-check-timeout-ms to allow independent configuration.

0.4.2

• Change the handle-postgres-permission-errors to generate a 500 code instead of a 403 when database permissions fail.

0.4.1

• Add a middleware namespace with ring middleware for catching Postgres permission errors and returning an http response.
• Add utility functions for pglogical support under the pglogical namespace.
• Add the puppetlabs/i18n library and externalize strings.

0.4.0

• Update java.jdbc dependency to 0.6.1. This version of java.jdbc has breaking changes.

0.3.1

• Convert PGObjects to their textual value in core/query - It is expected that users who don't want this functionality will either call jdbc/query directly or implement their own wrapper using the component functions called by core/query