[ https://issues.apache.org/jira/browse/HIVE-5206?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13758099#comment-13758099 ]
Edward Capriolo commented on HIVE-5206: --------------------------------------- Unifying the integral types is surely out of scope for this issue. However, We are making the justification for doing this for string, char, varchar. The same argument would hold for the integral types? I mean if someone was a fresh reader to the code and saw that all the string types work one way, but all the int types work another, what would they conclude is the "right" way? Are we thinking too patch driven development? Sure your end goal is to have this varchar support (and it is a good goal and this is a good feature) but who cares about all the existing stuff? Is it not sexy not worth touching? I am just thinking out loud here. But I notice we are starting to add many types, some like decimal are special purpose versions of other types. I want to make sure we are thinking about the big picture here. I am not sure we should just be adding a new type every time there is something we do not like about the existing type. What if someone wants nvarchar, or BigInteger, or BigDecimal? > Support parameterized primitive types > ------------------------------------- > > Key: HIVE-5206 > URL: https://issues.apache.org/jira/browse/HIVE-5206 > Project: Hive > Issue Type: Improvement > Components: Types > Reporter: Jason Dere > Assignee: Jason Dere > Attachments: HIVE-5206.1.patch, HIVE-5206.D12693.1.patch > > > Support for parameterized types is needed for char/varchar/decimal support. > This adds a type parameters value to the > PrimitiveTypeEntry/PrimitiveTypeInfo/PrimitiveObjectInspector objects. > NO PRECOMMIT TESTS - dependent on HIVE-5203/HIVE-5204 -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira