jarredhj0214 commented on PR #13562: URL: https://github.com/apache/gravitino/pull/13562#issuecomment-6072301876
@yuqi1129 Thanks! Addressed in 72e9936e4: restored the original methods as abstract, clarified the Javadoc contract, added the explicit default-`true` assertion and REST/Java examples, and updated the PR description with both Spark behavior changes. Python support remains a follow-up. All five targeted test classes and `:docs:build` pass. I reran the manager-level benchmark with explicit cache controls: 1,300 catalogs, three fresh test JVMs, 20 measured iterations per case per JVM (after warm-up). Mean times for `false` / `true`: - Cache cleared before every call: 0.110 / 86.368 ms (~783x). - Cache cleared only before each 20-call batch: 0.055 / 5.321 ms (~97x). - Fully warm cache: 0.056 / 0.499 ms (~8.9x). Cache-state assertions verified 0 wrappers before each cold call, 1,300 after the full listing, and 0 after the lightweight listing. These are new measurements using an in-memory store and test provider; I could not verify the original benchmark's cache handling. They do not measure end-to-end Spark startup. -- 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]
