1) It goes to the task log, which you can access through the Hadoop web ui. It does not go to your client console (unless your println is in the constructor or a couple other functions that are invoked on the client to set things up / check for correctness).
2) It's always a databag, and you do need to cast is as such. I wrote this blog post to explain a bit what group schema looks like and what you can and can't do with it: http://squarecog.wordpress.com/2010/05/11/group-operator-in-apache-pig/ On Sun, Oct 9, 2011 at 8:03 PM, Walter Chang <[email protected]>wrote: > Hi, > > I have two questions when using udf: > 1. if i do system.err.println , where does it print to ? is there a log > file > ? > > 2. if i did a group , what's the type of second column of the generated > table ? is it always databag ? So if i use it in my udf, i need to cast it > back to databag and use iterator to access each tuple, is it correct ? > > Thanks a lot, > > Weide >
