Details
-
Improvement
-
Status: Open
-
Minor
-
Resolution: Unresolved
-
1.7.3
-
None
-
None
Description
Summary:
If you want to use a grouping comparator with the "new" mapreduce API and Avro, you must provide an implementation of RawComparator compare(byte[] b1, int s1, int l1, byte[] b2, int s2, int l2).
In standard mapreduce, you don't have to do this. Hadoop does it for you behind the scenes by deserializing the writables and calling compare(x,y)
instead.
We should provide a subclass of AvroKeyComparator which does the deserialization and delegates to compare(x,y) to be consistent with Hadoop mapreduce.