Implementation Questions for improving Hive Alter parsing

[tsamaras]

I'm taking a crack at improving support for some of the quirks of Hive's alter commands and was hoping for some feedback on what the preferred approach is for solving a few problems.

Rather than ALTER COLUMN Hive uses CHANGE COLUMN with the following syntax

ALTER TABLE table_name [PARTITION partition_spec] CHANGE [COLUMN] col_old_name col_new_name column_type
  [COMMENT col_comment] [FIRST|AFTER column_name] [CASCADE|RESTRICT];
ALTER TABLE table_name CHANGE COLUMN column_name column_name data_type CONSTRAINT constraint_name constraint_spec

which means that every CHANGE is potentially a rename and data type update. Would it make more sense to define a new ChangeColumn expression or decompose the statement into an AlterColumn and RenameColumn expression that gets stitched back together in the generator? And the other thing i'm not sure if is what's the right way to deal with incompatible translations like a RenameColumn without a data type specified. My personal interest is in improving hive <-> spark2 translation so the implementation gets inherited either way but I'd like to make it as flexible as possible.