[ https://issues.apache.org/jira/browse/IMPALA-11491?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17827488#comment-17827488 ]
Daniel Becker edited comment on IMPALA-11491 at 3/15/24 12:27 PM: ------------------------------------------------------------------ In the backend the distinction between STRING and BINARY values is not handled in the normal PrimitiveType, both are TYPE_STRING. Differentiation is possible through AuxColumnType objects which are found in ColumnDescriptors. However, as far as I can see, nested fields do not have ColumnDescriptors. If/when we want to support nested BINARY fields we should consider removing AuxColumnType and using PrimitiveType::TYPE_BINARY for BINARY fields. In my opinion it is cleaner and and more straightforward, although it affects many places in the code. Alternatively we could make AuxColumnType a member of ColumnType so it is available everywhere but existing code that handles strings (and binaries where they should be handled identically) does not have to be changed. [~csringhofer] what do you think? was (Author: daniel.becker): In the backend the distinction between STRING and BINARY values is not handled in the normal PrimitiveType, both are TYPE_STRING. Differentiation is possible through AuxColumnType objects which are found in ColumnDescriptors. However, as far as I can see, nested fields do not have ColumnDescriptors. If/when we want to support nested BINARY fields we should consider removing AuxColumnType and using PrimitiveType::TYPE_BINARY for BINARY fields. In my opinion it is cleaner and and more straightforward, although it affects many places in the code. Alternatively we could make AuxColumnType a member of ColumnType so it is available everywhere but existing code that handles strings (and binaries where they should be handled identically) does not have to be changed. > Support BINARY nested in complex types in select list > ----------------------------------------------------- > > Key: IMPALA-11491 > URL: https://issues.apache.org/jira/browse/IMPALA-11491 > Project: IMPALA > Issue Type: Sub-task > Components: Backend, Frontend > Reporter: Csaba Ringhofer > Priority: Major > > The question is how the print BINARY types when the parent complex type is > printed to JSON. We can't rely on the existing Hive behavior, as it turned > out to be buggy: > HIVE-26454 -- This message was sent by Atlassian Jira (v8.20.10#820010) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-all-unsubscr...@impala.apache.org For additional commands, e-mail: issues-all-h...@impala.apache.org