i'm doing some experimenting, and noticed that schema resolution is not working quite as i had expected.
for example, i have two record schemas, one called R1 in namespace N1 and one called R2 in namespace N2. other than their names and namespaces, the records contain exactly the same field names and types. i create an avro data file containing records of type N1.R1. i can then read that file with a "specific" reader of type N2.R2. a generic reader reveals that the data file does indeed have the original N1.R1 schema, however. this is actually the behavior i want, as it would allow me to move and rename "specific" record classes in my java code without breaking serialized compatibility. however, it seems to violate the schema resolution rules found in the avro documentation: http://avro.apache.org/docs/1.4.0/spec.html#Schema+Resolution am i misunderstanding the documentation? is the behavior i'm seeing expected? when does a record name/namespace conflict actually cause an error to be thrown?
