You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@avro.apache.org by Johannes Schulte <jo...@gmail.com> on 2018/08/07 09:30:55 UTC
Mapreduce old vs new API: Ignoring non-avro files
Hi,
when using a directory with avro files with an Avro Hadoop InputFormat, do
I see it correctly that the old API implementation (in the mapred-package)
respects the property
avro.mapred.ignore.inputs.without.extension
(which defaults to true) using this code path
https://github.com/apache/avro/blob/cf2f30336efe0ecc3debc7bede86fde6d23f7c79/lang/java/mapred/src/main/java/org/apache/avro/mapred/AvroInputFormat.java#L54
while the new API (under mapreduce) that uses the term AvroKeyInputFormat
for the same use case ignores this property as it does not override the
default implementation of FileInputFormat?
https://github.com/apache/avro/blob/cf2f30336efe0ecc3debc7bede86fde6d23f7c79/lang/java/mapred/src/main/java/org/apache/avro/mapreduce/AvroKeyInputFormat.java#L40
Is this worth filing a pull request?
Cheers,
Johannes