pgjdbc currently maps the TIMESTAMP WITH TIME ZONE SQL type to the java.sql.Types.TIMESTAMP JDBC type. We are currently discussing whether we should instead map it to the java.sql.Types.TIMESTAMP_WITH_TIMEZONE JDBC for the JDBC 4.2 driver. We are unsure whether this would break Hibernate (assuming calling #getTimestamp would still work). Searching for references to TIMESTAMP_WITH_TIMEZONE we didn't find anything.

Is there a way to run the Hibernate test suite for PostgreSQL with a locally built SNAPSHOT version of the driver?

The PostgreSQL Dialect maps the JDBC TIMSTAMP Type to the DB timestamp type. Why do you want to map TIMESTAMP WITH TIMEZONE?

We do not want to change anything with the DB timestamp type.

For the "timestamptz" database type (TIMESTAMP WITH TIME ZONE) pgjdbc meta data currently reports java.sql.Types.TIMESTAMP as a JDBC type. We discuss reporting java.sql.Types.TIMESTAMP_WITH_TIMEZONE instead in this case. It currently doesn't seem to be mapped.

You can always extend the current PostgreSQL Hibernate Dialect and overwrite the JDBC Type for a given column type.

Yes, but we would like to know if this driver change would have an impact for existing Hibernate users.

When you say existing Hibernate users, you mean your the users that access you current application, right?

With existing Hibernate users I mean everyone who uses Hibernate with pgjdbc.

I think there was a misunderstanding.

I didn't say we plan on changing the current PosgreSQL Dialects to support TIMESTAMP WITH TIMEZONE.

I said that you should extend the Hibernate Dialect, defining your own Dialect which maps the JDBC Timsetamp to TIMESTAMP WITH TIMEZONE. So, it's just your application that will use your custom Dialect.

I as well believe there is a misunderstanding :-)

We are discussing to change the pgjdbc driver upstream. So any Hibernate user would get that change sooner or later. We would like to know if this change in the driver itself would negatively affect Hibernate users.

I'm going to ask this question internally to see what the other Hibernate Developers think of this Driver change.

From all the PostgreSQL references I found, it seems like everyone is suggesting to use TIMESTAMPZ instead of TIMESTAMP, so I guess this change is understandable.

From my perspective, this change will not just affect Hibernate users, but every JDBC framework that uses PostgreSQL. We also added a hibernate.jdbc.time_zone configuration property, so that our users can force all Timestamps to be saved in a given TimeZone, typically UTC.

I'm feeling anxious about this change since JDBC only offers a TIMESTAMP type. JDBC does not make a difference from TIMESTAMP and TIMESTAMP ith TIMEZONE, and not all DBs support this concept. For instance, MySQL makes the conversation automatically and saves all timestamps in UTC. But then, this is probably how TIMESTAMPZ works in PostgreSQL too.

I'm going to ask this question on Twitter as well to gather more feedback from our community as well.