rdblue commented on pull request #1843: URL: https://github.com/apache/iceberg/pull/1843#issuecomment-739012585
I think we should keep the scope of this small. There isn't a need to address how to embed path-based tables in SQL commands right now. However we choose to do that would probably be a different solution. Right now, we need to unblock multicatalog support in `IcebergSource` by having a way to pass the path through as a table identifier, and the `PathIdentifier` option works well for that. For the larger question about `ALTER`, `DROP`, etc., I think the short-term solution is to define a HadoopCatalog that can handle those SQL commands. In the long term, I've advocated that Spark should have some way of identifying and passing path identifiers to plugins. The problem here is that no one seems to know what the behavior or path-based tables in SQL is or should be. But I think that a similar `PathIdentifier` approach would work once that is addressed in the upstream community. Last, if we want to support identifiers with a catalog and a quoted path, we can do that any time by choosing when to interpret the catalog name as a quoted path. ---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
