[Zebra] Restrict schema definition for collection
-------------------------------------------------
Key: PIG-1269
URL: https://issues.apache.org/jira/browse/PIG-1269
Project: Pig
Issue Type: Bug
Reporter: Xuefu Zhang
Assignee: Xuefu Zhang
Fix For: 0.7.0
Attachments: zebra.0302
Currently Zebra grammar for schema definition for collection field allows many
types of definition. To reduce complexity and remove ambiguity, and more
importantly, to make the meta data more representative of the actual data
instances, the grammar rules need to be changed. Only a record type is allowed
and required for collection definition. Thus,
fieldName:collection(record(c1:int, c2:string)) is legal, while
fieldName:collection(c1:int, c2:string), fieldName:collection(f:record(c1:int,
c2:string)), fieldName:collection(c1:int), or feildName:collection(int) is
illegal.
This will have some impact on existing Zebra M/R programs or Pig scripts that
use Zebra. Schema acceptable in previous release now may become illegal because
of this change. This should be clearly documented.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
