If you really need speed, there are protocol buffers. Otherwise, I find the json with jackson works well.
On Mon, May 9, 2011 at 7:04 PM, Dhruv <[email protected]> wrote: > OK, going by the (in)efficiency arguments of JSON, I'll stick with my > original plan of writing a custom serializable class during the initial > phase of the project, which as Sean mentioned shouldn't be too much work > either. > > > On Mon, May 9, 2011 at 3:47 PM, Jake Mannix <[email protected]> wrote: > >> On Mon, May 9, 2011 at 12:03 PM, Dhruv <[email protected]> wrote: >> >> > Cloud 9 is an easy to use Hadoop MapReduce library by Jimmy Lin from the >> > University of Maryland using the Apache 2.0 license ( >> > http://www.umiacs.umd.edu/~jimmylin/cloud9/docs/). The library contains >> a >> > very convenient, lightweight JSON serializable class. One can use this >> > class >> > instead of rolling your own custom serializable objects and it could help >> > me >> > for the GSOC . >> > >> >> Jimmy's cloud9 library is primarily for educational use - it's free to >> include >> (license-wise), but not all of the classes in there are performance >> optimized, >> including this one. I don't think he's even benchmarked this against gson >> or jackson, but I can ask him (he is on leave from UMD and working at >> Twitter >> this year). >> >> Regardless, I agree with Sean - we pulled json *out* of mahout for a >> reason, >> let's not bring it back in. >> >> -jake >> >
