Closed tomwhite closed 13 years ago
I just updated the message as you suggested, which should be sufficient until HIVE-1850 can be used.
Here's the change to support compression (such as Snappy) in Avro files written by Hive.
I tried it out by running test_serializer.sql with the following lines added just before the "insert overwrite" line:
SET hive.exec.compress.output=true; SET avro.output.codec=snappy;
I then manually verified that the resulting Avro file was snappy compressed (by looking at the raw bytes), and that it could be read using Avro tools.
I also needed to copy the snappy-java jar to Hive's lib directory before running the test.
Unit tests pass and I managed to successfully create an Avro-backed Hive table and perform a simple query on it.