Github user marmbrus commented on the pull request: https://github.com/apache/spark/pull/12952#issuecomment-220116608 @liancheng, thanks for the thorough explanation. I think I understand the problem now and I agree with the proposed constraints. A few questions: > One tricky stuff in our current encoder design is that, it couples field reordering and Java object de/serialization. I think it's more reasonable to move field reordering into Project. Namely, make Project to be able to handle inner nested fields. Are you proposing that was always use a project to copy from one row to another such that the input is in the correct order for the deserializer's constructor? If so, I worry that that is too expensive. While its complicated, I think its pretty useful for us to be able to construct objects from an input row, independent of the ordering.
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. --- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org