grantatspothero commented on PR #18074: URL: https://github.com/apache/iceberg/pull/18074#issuecomment-5666111316
The caching approach looks more obviously correct than my previous attempt which was partially reverted: https://github.com/apache/iceberg/pull/15511 One risk is caching `ManifestFile` caches all fields, but the `ManifestFile` committed is not the same as the one in memory, see: `ManifestListWriter#prepare` which wraps each before writing. The cleanup methods only need a path and equals, so this is not a correctness concern today, but a cache of slightly invalid objects seems dangerous in the future. Did you explore changing the implementation of `protected abstract void cleanUncommitted(Set<ManifestFile> committed)` -> `protected abstract void cleanUncommitted(Set<String> committed);`? Then you can cache only the paths instead of the manifest files. This removes the need to cache invalid `ManifestFile` and you can cache only paths. -- 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]
