If jpa native query returns a custom db type (such as postgres "tsvector"), the jdbc type is OTHER(1111).
To correctly map the result, a custom type class must be provided to SQLQuery.addScalar(String columnAlias, Type type).
Without the above scalar type registration, query fails with a "org.hibernate.MappingException: No Dialect mapping for JDBC type: 1111", even if the custom type was registered in dialect via TypeContributions.contributeType.
It would be nice if the custom type was registered with a sql type (OTHER in this case), to check the PGObject and try to match a custom registered type.
This issue affets QueryDSL native jpa query functionality, as it internally uses SQLQuery.addScalar(String columnAlias), without any knowledge of the hibernate custom type.
PostgreSQL 9.6 (PostgreSQL Native Driver 9.4.1211), QueryDSL 4.1,4, Spring Boot 1.4.0