HeartSaVioR commented on pull request #28026: URL: https://github.com/apache/spark/pull/28026#issuecomment-624460580
I think we agreed at least two things: 1. current state of create table brings "more" confusions than before (Spark 2.x) 2. unifying create table syntax is the way to go and one more thing seems to be true as of now, unifying create table syntax should have more discussion (and maybe with vote?) to go forward and it seems to be tough to happen in Spark 3.0.0 release plan. Thinking ideally, current state sounds to me as "unstable" / "experimental", and we cannot make an "experimental" change on create table syntax, as the feature has been provided for a long time and it has been used in production. Breaking backward compatibility should have huge benefits over the previous one (that's the rubric AFAIK), and I'm wondering current state (including band-aid fixes on running out of time) would provide enough benefits. I understand there has been lots of efforts on the data source table (non-Hive I meant) and we would like to enable it by default, but I don't think it worths to bring backward incompatibility just for removing `USING ...`. While I agree about the importance of changing "default", I also concern about changing "default" as well. That's the same importance. Even if we revert the change, we can still encourage to use USING clause whenever possible (with the promise of unified create table syntax in near future if we want), but if we intend to leave the state as it is (or try to deal with band-aid fixes) we can't prevent existing query to fail or work as unintentional way. They will have to make some changes on their query, or even worse, have to take workarounds. And once unified create table syntax is in, they might have to make some changes on their query again. Why not postponing all the changes (including the changes we did) until we have a good shape of new approach to let end users deal with it only once in near future? ---------------------------------------------------------------- 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]
