prosgarz35 commented on PR #3228: URL: https://github.com/apache/james-project/pull/3228#issuecomment-5971620383
Hi @chibenwa, Thanks for the healthy pushback — you were totally right to question this! I went ahead and ran some profiling across the test suite to see the real impact: Pure in-memory I/O with JimFS is indeed crazy fast in micro-benchmarks (~90–100× faster, roughly 70 ms vs 6.7 s for 1k raw cycles), but actual disk-bound tests make up less than 0.3% of James' overall test time. The full CI run sits around ~1h 23m, and the vast majority of that is taken up by Testcontainers, Guice/Spring wiring, Netty sockets, and the distributed backends (Postgres, Cassandra, OpenSearch, RabbitMQ/Pulsar). Even if we migrated FileBlobStoreDAO and adjacent modules, we'd only shave off maybe ~15–25 seconds at best — basically lost in CI run-to-run noise (±3–5 mins). All in all, pulling in a new dependency and maintaining a custom test extension for a few seconds saved definitely falls into classic premature optimization territory. Not worth the extra maintenance overhead. Closing this PR with no regrets — thanks again for taking the time to share your perspective and keeping things pragmatic! -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
