The existing json format is a leftover of our initial and now replaced native format. As such it is tweaked to work best with Hadoop, namely newlines and tabs have to be escaped everywhere. Existing JSON data usually needs some reformatting to be read by rmr without a custom parser and that's inconvenient. To be able to implement a parser on top of RJSONIO, our current choice of RJSON parsing, we need RJSONIO to have a connection interface and the RJSONIO author has ongoing work towards that goal. As soon as it is read we will start working on a more general purpose JSON reader and writer. The old one will be renamed json-rmr or json-legacy or something.
The existing json format is a leftover of our initial and now replaced native format. As such it is tweaked to work best with Hadoop, namely newlines and tabs have to be escaped everywhere. Existing JSON data usually needs some reformatting to be read by rmr without a custom parser and that's inconvenient. To be able to implement a parser on top of RJSONIO, our current choice of RJSON parsing, we need RJSONIO to have a connection interface and the RJSONIO author has ongoing work towards that goal. As soon as it is read we will start working on a more general purpose JSON reader and writer. The old one will be renamed json-rmr or json-legacy or something.