[
https://issues.apache.org/jira/browse/DRILL-836?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Aditya Kishore updated DRILL-836:
---------------------------------
Attachment: DRILL-836-Drill-needs-to-return-complex-types-e.g.-m.patch
+ With the new patch, we return VARCHAR instead of VARBINARY.
+ Added test case.
+ Minor code refactoring.
> Drill needs to return complex types (e.g., map and array) as a JSON string
> --------------------------------------------------------------------------
>
> Key: DRILL-836
> URL: https://issues.apache.org/jira/browse/DRILL-836
> Project: Apache Drill
> Issue Type: Bug
> Components: Execution - Operators, Execution - RPC
> Affects Versions: 1.0.0-BETA1
> Reporter: George Chow
> Assignee: Aditya Kishore
> Priority: Blocker
> Fix For: 1.0.0-BETA1
>
> Attachments:
> DRILL-836-Drill-needs-to-return-complex-types-e.g.-m.patch,
> DRILL-836-Drill-needs-to-return-complex-types-e.g.-m.patch
>
>
> Drill needs to help users understand the available columns in a given HBase
> table's column-family. One way to do this is to implement the DESCRIBE
> command for a column-family. The purpose of this is to let Drill do the
> column sampling. The mandatory LIMIT clause specifies the sample size so as
> to throttle the number of columns returned.
> ==
> Instead of the above narrow proposal, a general mechanism to return complex
> types in a JSON string so as to allow client tools (such as the ODBC driver)
> to operate on these complex types.
> Returning a map as JSON would provide transparency into HBase column-families
> and allow the ODBC driver to help surface the names of columns within a
> column-family.
> Returning an array as JSON would provide transparency into CSV files since
> that conveys the number of columns to the ODBC driver.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
