adiranjan25 commented on issue #18921: URL: https://github.com/apache/iceberg/issues/18921#issuecomment-6888658634
Hi @ashishkumar50 and Iceberg maintainers, I'm interested in contributing to this improvement, particularly the proposed separation of canonical metadata URIs from engine-specific physical storage URIs. From an implementation perspective, I see several areas worth validating: 1. A configurable `CanonicalUriFileIO` that delegates to an existing FileIO while resolving canonical URIs into protocol-specific locations. 2. Preservation of canonical locations returned through `InputFile` and `OutputFile`, so physical access URIs do not unintentionally leak into persisted Iceberg metadata. 3. Consistent behavior for reads, writes, deletes, and bulk operations. 4. Backward compatibility with existing absolute-path tables and consideration of Format V4 relative paths. 5. Tests demonstrating access to the same logical files through multiple protocol mappings without requiring a specific storage provider. I would be happy to develop a focused proof of concept with unit and integration tests. Before starting, could maintainers confirm whether a generic FileIO wrapper is the preferred architectural direction, or whether this capability should be handled by existing FileIO implementations or engine-specific configuration? I would also appreciate guidance on the appropriate module and expected compatibility requirements. Thanks! -- 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]
