Hello Everyone, Thank you for your feedback on this pull request (mostly via GitHub comments). I’ve recently updated it, and I think I’ve addressed most of the issues. https://github.com/apache/calcite/pull/3668
There’s still an open discussion around the coupling with JTS. I’m fairly comfortable with the current state (though perhaps a little too eager to move on), but I’d like to get your perspectives before merging. If we decide further decoupling is needed, it would help if someone could outline or suggest a clean, elegant approach. For instance, the Parameter.sqlType annotation was a significant step forward, and I wouldn’t have arrived at this solution without Julian’s guidance. Similarly, I think we’d benefit from a deeper discussion to outline the architecture for a fully decoupled solution. Thanks in advance for your thoughts, Bertil > On 8 Jun 2025, at 11:45, Bertil Chapuis <bchap...@gmail.com> wrote: > > Hello Everyone, > > CALCITE-6239 [1] and CALCITE-6263 [2] were stalled for quite a while (sorry > about that). They are about adding a PostGIS dialect to Calcite and loosening > the coupling of RelDataTypeFactoryImpl to Java Topology Suite (JTS). I have > recently been able to address most of the comments and would love to get some > feedback. I believe most of the original concerns have now been addressed, > but given the long delay, it probably makes sense to have a new round of > reviews. > > Please let me know what you think in the JIRA issues or in the PR [3], which > now includes an updated description. As the changes are quite significant and > affect the handling of GEOMETRY types, I would feel more comfortable > receiving feedback before merging. > > Thanks a lot for your help, and best regards, > > Bertil > > [1] https://issues.apache.org/jira/browse/CALCITE-6239 > [2] https://issues.apache.org/jira/browse/CALCITE-6263 > [3] https://github.com/apache/calcite/pull/3668 >
