All, We are working on a native (aka C++) ORC reader and writer. For now we are working on it over at my old github - https://github.com/hortonworks/orc . First of all, I wanted to let everyone know it is happening and invite others to give feedback or help. You can see the API for the reader in the src/orc directory.
It leads to an interesting question. I'd like to contribute it back to Hive to keep the two implementations (Java and C++) together, but that would mean adding an optional C++ module to Hive, which is currently all Java. The other option is to take the native ORC reader to Apache incubator as a new project and eventually pull the Java one along with it. I'm very interested in the Hive development communities opinion. Thanks, Owen
