Description
The option should define a sequence of bytes between two consecutive json records. Currently the separator is detected automatically by hadoop library:
https://github.com/apache/hadoop/blob/trunk/hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/util/LineReader.java#L185-L254
The method is able to recognize only \r, \n and \r\n in UTF-8 encoding. It doesn't work in the cases if encoding of input stream is different from UTF-8. The option should allow to users explicitly set separator/delimiter of json records.