Jihoon Son created TAJO-1555:
--------------------------------
Summary: Duplicated code cleanup
Key: TAJO-1555
URL: https://issues.apache.org/jira/browse/TAJO-1555
Project: Tajo
Issue Type: Task
Components: function/udf
Reporter: Jihoon Son
I'm working on supporting Python UDF at TAJO-1344. This is still a prototype,
and has some problems. One of the problems is related to
serialization/deserialization protocol. For easy implementation, I simply used
CSV format to serialize/deserialize tuples. To do so, I copied some bunch of
codes from the tajo-storage package. This will incur a maintenance issue in
addition to the problem of low performance.
To cleanup the duplicated codes, I think that we should use a well-known
serialization/deserialization protocol such as protocol buffers.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
