reductionista edited a comment on pull request #554: URL: https://github.com/apache/madlib/pull/554#issuecomment-790936348
> The rule is to just give a table name without any schema. We append the madlib schema and create the table there. If you want > to delete a function, there is `delete_custom_function` with the same rule (no schema in the table name). This will actually drop > the table if you remove every custom function therein. If you don't want to delete the functions one-by-one and want to > completely get rid of the table right away, then you'll have to do it manually by using the madlib schema. If that seems > confusing, we can add a new function to drop the whole table as well. I still think the ideal behavior would be to let both the admin and the user pass in any path to any table in any schema they want, either fully qualified or relative to their search_path. I think this would be simpler and probably less risky, as well as allowing the admin more flexibility in where they want to have the table stored. But if it's easier to leave it the way it is for this release, then I'm fine with that... as long as we're sure there aren't any loopholes. I think the main thing we need to be sure of, in either case, is that the table we check permissions on is the same one we read from--so let's make sure any modification of what they pass in happens before the check rather than after. ---------------------------------------------------------------- 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: us...@infra.apache.org
