I ask because I was re-looking at https://issues.apache.org/jira/browse/HIVE-4734 and how to support newer data types (Decimal, Varchar, Char)
In this qtest, we write using a schema containing a decimal having precision 5 and scale 2 (decimal(5,2)) , but when reading the same data we set the read schema having a decimal data type of decimal(4,1) . This test produces correct results* (because the AvroSerde uses the file schema and not the reader schema to read the decimal type)* but is the schema evolution for the decimal type valid? Avro schema evolution spec is not very clear what is a valid evolution rule for decimal types. From the Avro spec ( http://avro.apache.org/docs/1.8.1/spec.html#Decimal) I see the following comment *For the purposes of schema resolution, two schemas that > are decimal logical types match if their scales and precisions match.* Is this effectively saying that reading schema for decimal type should match in precision and scale with the writer schema?