voonhous commented on code in PR #18837: URL: https://github.com/apache/hudi/pull/18837#discussion_r3466712962
########## docker/hoodie/hadoop/trinobase/scripts/trino.sh: ########## Review Comment: The trino docker modules (`trinobase`/`trinocoordinator`/`trinoworker`) existed to package the old external connector for the docker demo: `trinobase`'s `package`-phase antrun step copies `packaging/hudi-trino-bundle/target/hudi-trino-bundle-*.jar` into the image. RFC-105 removes that bundle — the connector now ships as the regular `hudi-trino` artifact consumed from the Trino-side `trino-hudi` shim — so the copy has no source jar. And it isn't skippable: `-Pintegration-tests` pulls `docker/hoodie/hadoop` into the reactor, which builds those modules, and the antrun copy runs at `package` (not gated by `docker.build.skip`, which only skips the `dockerfile-maven-plugin` image build). So with the bundle gone, any `-Pintegration-tests` build fails at `trinobase`'s package phase on a missing source jar. Rather than leave that dangling reference, I dropped the three modules along with their compose/pom/bundle references here. Connector coverage moved to the standard Trino connector test suite in `hudi-trino-plugin` — `TestHudiConnectorTest`, `TestHudiSmokeTest`, `TestHudiMinioConnectorSmokeTest`, `TestHudiSharedMetastore`, plus the file-operation and custom-merger E2E tests — which run against an in-process Trino query runner instead of the docker-compose demo. On validating against a public Trino release + quickstart — agreed that's worth having. Under the two-repo model the connector also ships to users through Trino's own release via the `trino-hudi` shim, so that validation fits naturally on the Trino side too. The stack removed here can't be revived as-is, though: it was coupled to the now-removed bundle and pinned to stale versions (Trino 368, Hadoop 2.8.4, Hive 2.3.3, Presto 0.271, on the deprecated `dockerfile-maven-plugin`), none of which works with the migrated connector (Trino SPI 481 / JDK 25). So a quickstart in the Hudi repo would mean a fresh docker setup against a current Trino, Let me double check and create an issue/task to track this. Does that sound reasonable? -- 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]
