It's not JSON, per se, but data formats like smile (
http://en.wikipedia.org/wiki/Smile_%28data_interchange_format%29) provide
support for markers that can't be confused with content and also provide
reasonably similar ergonomics.
—
p...@mult.ifario.us | Multifarious, Inc. | http://mult.ifario.us/
Moving over to the dev list, as this isn't a user-scope issue.
I just ran into this issue with the missing saveAsTestFile, and here's a
little additional information:
- Code ported from 0.9.1 up to 1.0.0; works with local[n] in both cases.
- Driver built as an uberjar via Maven.
- Deployed to sma
I agree that it will be good to see more algorithms added to the MLlib
universe, although this does bring to mind a couple of comments:
- MLlib as Mahout.next would be a unfortunate. There are some gems in
Mahout, but there are also lots of rocks. Setting a minimal bar of
working, correctly impl
I'd hazard that this is a generic issue. The "store" is in the context of
the driver code, not the worker code, and that's why Spark is trying to
send it off to a worker for execution. It's not serializable (and
shouldn't be...), so that fails.
Try making a Scala object that lives on the worker
