amrishlal edited a comment on pull request #7873: URL: https://github.com/apache/pinot/pull/7873#issuecomment-990367924
> can we decide on the delimiter right now? @siddharthteotia > > I really prefer not to overload . in the identifier. SQL standard is tableName.columnName. Can we something like postgres person -> age or like spark person:age > > we want to support unnest, join etc and this will add a lot of complexity to maintain backward compatibility. @kishoreg We can change the delimiter if you really want, but this isn't really overloading. We are just applying the standard dot notation to go to finer level of detail, i.e going from table to column to JSON elements within the column values. Please note that dot notation will only work for columns of type JSON. Oracle JSON database already uses this notation for JSON documents, so this isn't exactly without precedence. I fail to see why postgres went with "->" (at least to me it seems like an arbitrary choice). A secondary issue here is that delimiters such as "->" or ":" in identifiers are not supported by calcite (but ".", "[", and "]" are), so that allows us to suffix JSON path expressions behind column name. I had discussed this issue a bit with folks on calcite side as well (and my understanding is that they are open to extending the notation to include json path expressions in future). In short, I think we should be ok with using "." here. Happy to discuss further if you wish. -- 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]
