[CRUNCH-667] Documentation should mention that cogrouping requires a common type family

• Type: Improvement
• Status: Open
• Priority: Minor
• Resolution: Unresolved
• Component/s: Core
• Assignee: Josh Wills
• Reporter: Andrew Olson
• Created: 13/Mar/18 18:33
• Updated: 13/Mar/18 18:34

Description

Attempting to cogroup an Avro-based table with a Writable-based table produces an exception like:

java.lang.ClassCastException: org.apache.crunch.types.writable.WritableType cannot be cast
   to org.apache.crunch.types.avro.AvroType
at org.apache.crunch.types.avro.Avros.collections(Avros.java:409)
at org.apache.crunch.types.avro.AvroTypeFamily.collections(AvroTypeFamily.java:106)
at org.apache.crunch.lib.Cogroup.cogroup(Cogroup.java:63)
at org.apache.crunch.lib.Cogroup.cogroup(Cogroup.java:46)
at org.apache.crunch.impl.dist.collect.PTableBase.cogroup(PTableBase.java:176)

This technical limitation makes sense, and the exception message is relatively self-explanatory (at least for anyone who is familiar with Crunch concepts), so I don't think there's really a bug to be fixed. However both the documentation in the code javadoc (Cogroup class, Set#difference, and PTable#cogroup) and the user guide section about Cogroups (https://crunch.apache.org/user-guide.html#cogroups) could be updated to clarify that all tables involved in a cogroup operation must have the same type family.