Hibernate Query Timeout Configuration Missing Cross-Version Cancellation Guarantee

Hibernate Query.setTimeout() establishes a JDBC statement timeout, but the framework does not guarantee cancellation of the underlying query when the timeout expires. This behavior creates ambiguity for applications that rely on timely query termination, especially when pagination via setMaxResults and setFirstResult is involved. The timeout applies at the statement level, and Hibernate's session management does not automatically propagate cancellation to the JDBC driver, potentially leaving threads active after the timeout period. Furthermore, exception handling differs across Hibernate major versions: version 5 raises QueryTimeoutException directly, while version 6 may wrap the exception in PersistenceException or treat it as a runtime exception, affecting upstream error handling strategies. The interaction between pagination boundaries and timeout enforcement also raises questions about whether a timeout applied to a paginated query respects the configured page size or continues fetching beyond it.

Under what conditions does Hibernate.cancel() effectively terminate a timed-out query across supported versions? How should applications reconcile version-specific exception types when implementing cross-platform timeout logic? Does configuring setMaxResults in conjunction with setTimeout() guarantee that excess results are discarded once the timeout fires, or may the driver continue returning rows beyond the page boundary?