Hi Ryan, I think the Python impl does the right thing.
You say that Java currently works fine with "record". Does it work because you tried with Java 14- (i.e. no support for record types - https://docs.oracle.com/en/java/javase/14/language/records.html) or because the special name "record" is mangled to "record$" ? In any case your example in the ticket {"type":"record","name":"LinkedList","fields":[ {"name":"value","type":"int}, {"name":"next","type":["null",{"type":"LinkedList"}]}]}" convinced me that "record", "enum" and "fixed" should be treated as the primitive types. Martin On Fri, Feb 4, 2022 at 1:13 PM Ryan Skraba <r...@skraba.com> wrote: > Hello! I created the JIRA AVRO-3370[1] that demonstrates two > different behaviours between Java and python with respect to using > complex types (such as "record" as a name in a named type). > > This is specifically an issue because Flink is generating these types > of records (in the Java world), and they are subsequently not > interoperable with Python users.[2] > > There's probably two fixes to be done here: > > - Better define the behaviour so all language SDKs are consistent, and > - Contribute an upstream fix to Flink so that it's compatible with python. > > What do you think is the right thing to do? Should we be able to > define a record named record (the Java behaviour) or should the spec > be stricter about using types like names (the Python behaviour)? > > All my best, Ryan > > [1]: https://issues.apache.org/jira/browse/AVRO-3370 > [2]: > https://github.com/apache/flink/blob/release-1.14.3/flink-formats/flink-avro/src/main/java/org/apache/flink/formats/avro/typeutils/AvroSchemaConverter.java#L307 >
