JDK 27 Release Schedule Formalized
Oracle Chief Architect Mark Reinhold has formally proposed the JDK 27 release timeline, locking in the dates that will govern the next six-month Java release cycle. The schedule targets Rampdown Phase One on June 4, 2026, followed by Rampdown Phase Two on July 16. Initial and Final Release Candidates are set for August 6 and August 20 respectively, with General Availability on September 14, 2026.
Build 17 of the early-access releases is already available, incorporating recent bug fixes. The most significant JEP already targeted is JEP 527 — Post-Quantum Hybrid Key Exchange for TLS 1.3, which will add quantum-resistant cryptographic handshakes to Java's TLS stack using a hybrid approach that combines classical and post-quantum algorithms. JDK 27 will be a non-LTS release, continuing the cadence established since JDK 21.
Developers building long-lived applications should take note of the GA date when planning dependency upgrades and preview feature evaluations, as preview features from JDK 26 that progress to candidate or final status in JDK 27 will lock their APIs.
Read more — InfoQ
Hibernate ORM 7.3.0
Hibernate ORM 7.3.0 introduces a KeyType enumeration that enables overloaded find() methods based on natural IDs, giving developers a type-safe way to distinguish between primary key and natural key lookups without relying on method overloading ambiguity. The new @NaturalIdClass annotation models non-aggregated composite natural IDs, aligning Hibernate's natural ID handling with Jakarta Persistence standards and enabling more expressive entity loading strategies.
The release also brings a new bulkWrite() method to the data operations API and additional refinements to the session-scoped natural ID resolution cache, improving performance for applications that load entities by business keys frequently. Applications using Hibernate 7.x can adopt these improvements by upgrading the hibernate-core dependency; the natural ID API changes are backward compatible with existing @NaturalId annotations.
Read more — InfoQ
LangChain4j 1.13.0
LangChain4j 1.13.0 ships with two classes central to building production-grade agentic systems: RecoverabilityIT and PendingResponse. Together they provide a framework for persistent, recoverable execution states that allow long-running AI workflows to survive restarts and handle intermittent failures without losing progress — a critical gap for agent orchestration in production Java environments.
The new ClassPathSkillLoader class resolves skill definitions from the classpath, simplifying deployment of AI agent skills packaged inside standard Java JARs. Additionally, a HibernateContentRetriever enables data retrieval through Hibernate Query Language queries, allowing Java AI applications to pull structured data from relational databases as retrieval-augmented generation (RAG) context without custom integration code.
These additions continue LangChain4j's focus on bridging the Java ecosystem — Spring, Hibernate, standard JVM tools — with LLM-based application patterns. Developers working with Quarkus or Spring Boot can consume these features through the existing LangChain4j integrations for both frameworks.
Read more — InfoQ
Keycloak 26.6.0
Keycloak 26.6.0 delivers full RFC 7523 support for JSON Web Token Profile for OAuth 2.0 Client Authentication, enabling JWT assertion-based token requests where clients authenticate using signed JWTs rather than client secrets. This is a widely-requested enterprise feature for service-to-service authentication scenarios where secret rotation is operationally difficult.
An experimental OAuth Client ID Metadata Document capability has been added for Model Context Protocol compatibility, allowing MCP-based AI agents to authenticate with Keycloak-protected resources using a standardized discovery mechanism. The release also delivers complete integration of the Keycloak Test Framework based on JUnit 6, replacing the older test harness and making it significantly easier to write integration tests for custom Keycloak extensions, themes, and authentication flows.
Developers maintaining custom SPIs or Keycloak extensions should review the test framework migration guide, as the JUnit 6-based approach changes how test containers are configured. The MCP OAuth experimental support signals Keycloak's direction toward becoming the identity provider of choice in agentic AI architectures.
Read more — InfoQ
Google ADK for Java 1.1.0
Google's Agent Development Kit for Java reaches version 1.1.0, building on the 1.0 GA release with additions focused on chat-centric workflows and broader model support. New ChatCompletionsRequest and ChatCompletionsResponse classes serve as data transfer objects for chat-style interactions, providing a clean API boundary between agent orchestration logic and the underlying model calls.
Gemma model support has been added to LlmRegistry, allowing developers to run Google's open-weight Gemma models within ADK-orchestrated agent workflows alongside the existing Gemini integrations. A new EventData abstract class facilitates tracing and observability management by providing a structured base for tracking agent execution events, which is essential for debugging complex multi-step agent flows.
The 1.1.0 release is a minor feature drop on top of the recently stable 1.0 baseline, and existing ADK for Java 1.0 projects can upgrade without breaking changes. The addition of Gemma support is particularly useful for teams that need on-premise or cost-controlled model inference within their Java agent pipelines.
Read more — InfoQ
Junie CLI: Auto-Detect JetBrains IDEs
JetBrains announced that their Junie CLI can now automatically detect and connect to any JetBrains IDE installed on a developer's workstation. This beta capability bridges the gap between terminal-based AI workflows and IDE-native features, allowing Junie CLI to understand project context, run tests automatically, perform precise refactoring, and execute semantic code navigation across complex multi-module projects.
Previously, Junie CLI operated in a standalone mode disconnected from the IDE's indexing and language understanding. With IDE auto-detection, commands issued from the terminal now benefit from the same deep project model that powers IntelliJ IDEA's in-editor analysis — meaning refactoring operations are symbol-aware and test execution reflects the IDE's run configurations rather than a generic build command.
This feature is available in beta and requires a compatible JetBrains IDE installation alongside the Junie CLI. Teams using JetBrains IDEs for Java, Kotlin, or Spring development can benefit from a more coherent workflow where terminal-driven AI automation and IDE assistance share the same project understanding.
Read more — InfoQ
Links & Sources
You May Also Enjoy
