I mentioned it in my previous post, but what are the lists thoughts on using a different versioning strategy for the driver in a similar manner to other OS projects.
My suggestion would be to: - freeze the current release at v1.x for partial JDBC4 support - archive the pre JDBC4 versions completely - start a new version 2, for example incorporating the major NIO and XA changes previously discussed running on the current JDK incorporating stubs or complete functions for the latest JDBC spec features as needed, etc and supporting the latest versions of Postgres as appropriate
Work would progress with v2 with only significant bug fixes to then be propagated back to v1.
Hopefully this reaches some compromise for the stability issues with existing users who can remain on v1 with newer projects migrating to the v2.
At some point in the future I'd envisage a new version being released for example with the release of JDK8 or JDBC5 as required..