davidradl commented on code in PR #839: URL: https://github.com/apache/flink-web/pull/839#discussion_r3176452882
########## docs/content/posts/2026-04-13-community-update.md: ########## @@ -0,0 +1,166 @@ +--- +authors: +- jlpayne72: null + name: Julian Payne +- davidrad: null + name: David Radley +date: "2026-04-13T08:00:00Z" +excerpt: Flink Community update for April 2026 +title: Flink Community update for April 2026 +aliases: +- /news/2026/04/13/community-update.html +--- + +This is the monthly Flink Community update for April 2026. We scour the latest updates from the Flink community so you don't have to! + +<!--more--> + +We'll dive into a few hot topics and FLIPs (Flink Improvement Proposals) before providing the usual almanac-style listing of technical updates, user-facing updates, and governance and community. + + +<!-- TOC --> + * [Hot Topics](#hot-topics) + * [Developer/Technical Updates](#developertechnical-updates) + * [User-Facing Updates](#user-facing-updates) + * [Governance and Community](#governance-and-community) +<!-- TOC --> + +## Hot Topics + +### Apache Flink 2.3 Release Draws Nearer + +Four months after the [release of Flink 2.2](https://flink.apache.org/2025/12/04/apache-flink-2.2.0-advancing-real-time-data--ai-and-empowering-stream-processing-for-the-ai-era/) (December 4th 2025), we are hotly anticipating the release of Flink 2.3. +The code freeze for Apache Flink 2.3 is planned for April 14th and at time of writing we anticipate the version to be launched toward the end of April 2026. +Some of the main features in this release will include important updates to Materialized Tables for Flink SQL, OTel gRPC exporter and watermark alignment for backlogged jobs in Runtime, and in Connectors, an update to the S3 FileSink connector. +Stay tuned to this blog for the release announcement with full details. +We will also restructure the Flink documentation to make it easier to navigate for new and existing users. +You can see the full release scope [here](https://cwiki.apache.org/confluence/display/FLINK/2.3+Release). + +### Flink CDC 3.6.0 Release + +Flink CDC is used to capture and stream real-time changes from databases (inserts, updates, and deletes) as they happen. +[Version 3.6.0](https://flink.apache.org/2026/03/30/apache-flink-cdc-3.6.0-release-announcement/) extends Flink version support to 1.20.x and 2.2.x, upgrades JDK version to 11, introduces new [Oracle Source](https://nightlies.apache.org/flink/flink-cdc-docs-release-3.6/docs/connectors/pipeline-connectors/oracle/) and [Apache Hudi Sink](https://nightlies.apache.org/flink/flink-cdc-docs-release-3.6/docs/connectors/pipeline-connectors/hudi/) Pipeline connectors and adds Lenient mode schema evolution support for [Fluss Pipeline connector](https://nightlies.apache.org/flink/flink-cdc-docs-master/docs/connectors/pipeline-connectors/fluss/). +It also introduces [PostgreSQL Schema Evolution](https://nightlies.apache.org/flink/flink-cdc-docs-release-3.6/docs/connectors/pipeline-connectors/postgres/) support, enhancing Schema Evolution capabilities for better multi-table synchronization scenarios and table name mapping flexibility, and strengthening the Transform framework with VARIANT type and JSON parsing support. + + +## Developer/Technical Updates + +### Table API & SQL + +Continued work on Materialized Tables including `START_MODE` clause parser support ([FLINK-39304](https://issues.apache.org/jira/browse/FLINK-39304)), interval reuse for `FRESHNESS` and `START MODE`, and `CREATE OR ALTER` now correctly respects existing table schemas ([FLINK-39284](https://issues.apache.org/jira/browse/FLINK-39284)). `TO_CHANGELOG` retract/upsert stream conversion was also added. + +`BITMAP` Type support was introduced end-to-end: the core data type ([FLINK-38852](https://issues.apache.org/jira/browse/FLINK-38852), [FLINK-39185](https://issues.apache.org/jira/browse/FLINK-39185)), scalar functions ([FLINK-39186](https://issues.apache.org/jira/browse/FLINK-39186)), aggregate functions ([FLINK-39187](https://issues.apache.org/jira/browse/FLINK-39187)), and documentation ([FLINK-39188](https://issues.apache.org/jira/browse/FLINK-39188)). + +Join Operations saw cascaded delta join support merged ([FLINK-39233](https://issues.apache.org/jira/browse/FLINK-39233)), lookup join after delta join ([FLINK-39174](https://issues.apache.org/jira/browse/FLINK-39174)), and immutable column sink mode traits ([FLINK-39287](https://issues.apache.org/jira/browse/FLINK-39287)). This removes restrictions, allowing the delta join to be used in more scenarios. + +Code generation and correctness work included: NDU analyzer improvements for non-deterministic function detection ([FLINK-39313](https://issues.apache.org/jira/browse/FLINK-39313)), duplicate function instance elimination ([FLINK-39094](https://issues.apache.org/jira/browse/FLINK-39094)), filter optimisation for upsert key groups ([FLINK-39314](https://issues.apache.org/jira/browse/FLINK-39314)), a UDF lambda expression argument fix ([FLINK-39273](https://issues.apache.org/jira/browse/FLINK-39273)), and JSON field name code generation fixes backported across multiple releases ([FLINK-39355](https://issues.apache.org/jira/browse/FLINK-39355)). + +### Runtime & Execution + +The end of the month saw rescale enhancements with the rescale history recording statistics ([FLINK-38343](https://issues.apache.org/jira/browse/FLINK-38343)) and rescale information ([FLINK-38342](https://issues.apache.org/jira/browse/FLINK-38342)). A configurable `ExecutionGraph` cache TTL was introduced ([FLINK-39016](https://issues.apache.org/jira/browse/FLINK-39016)), and the Flink Web UI was updated to surface rescale and configuration data for adaptive scheduler jobs ([FLINK-38901](https://issues.apache.org/jira/browse/FLINK-38901)). + +### Checkpointing & State + +Unaligned checkpoint recovery gained a new configuration option ([FLINK-38541](https://issues.apache.org/jira/browse/FLINK-38541)), giving operators finer control over recovery behaviour under failure conditions. Session-mode HA recovery has been enhanced with [FLINK-38975](https://issues.apache.org/jira/browse/FLINK-38975), introducing new Application stores for running and terminated applications. + +### Metrics & Observability + +gRPC batch export support was added to the metrics pipeline ([FLINK-39126](https://issues.apache.org/jira/browse/FLINK-39126)), providing a more robust export using gRPC for jobs with large numbers of operators or tasks. + +### Python & Multi-language Support + +Python users gained support for multiple compression formats in `python.files` ([FLINK-39260](https://issues.apache.org/jira/browse/FLINK-39260)), expanding deployment flexibility for Python UDF bundles. + +## User-Facing Updates Review Comment: @rmoff I think what we are doing each month is fitting the titles around the content that we have, rather than keeping a consistent format. I am ok with that, as we then do not have empty of scant sections. I do not have strong views on this. Maybe after more blog updates we will settle on meaningful sections, lets see how the content goes. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
