Hunter Logan created HIVE-23861:
-----------------------------------
Summary: Beeline Output Formats should parse underlying resultset
themselves
Key: HIVE-23861
URL: https://issues.apache.org/jira/browse/HIVE-23861
Project: Hive
Issue Type: Improvement
Components: Beeline
Reporter: Hunter Logan
Currently *Outputformat types in Beeline are passed a Rows object. This rows
object parses a resultset
([https://docs.oracle.com/javase/7/docs/api/java/sql/ResultSet.html]) into
strings and stored in Rows.values. The *Outputformat types do not have access
to the underlying resultset just the values array and the metadata.
The benefit to this current design is that every output format is forced to
output the data the identically while only changing the aesthetics. The
downside to this is that it restricts _actually_ marshaling the data into a new
type. For example Outputformats can't support an SQL Array type because they
have no way to find the types within the Array, so Arrays are printed as a
string, e.g. "["1","2"]"
I am proposing that Outputformats be refactored to be passed the resultset
directly. For starters this allows us to add support for Arrays but also allows
for Outputformats that actually output the results into a more useable format,
e.g. AvroOutputformat, ParquetOutputformat.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
