Wouldn't that add extra overhead to the process? I mean to do an extra FOREACH ... GENERATE won't cause an extra Mapreduce job to be generated?
Renato M. 2010/7/25 Jai Krishna <[email protected]> > Ok. That helps. > So for this situation, we need not write a new UDF; we can just use > FOREACH...GENERATE to rearrange the tuple columns. > > -RJK > > > On 7/23/10 1:13 PM, "Harsh J" <[email protected]> wrote: > > Yes, that _will_ guarantee that the ordering is what you've specified. > > On Fri, Jul 23, 2010 at 11:33 AM, Jai Krishna <[email protected]> wrote: > > So a question on a related note, is there any correlation between the way > the tuple was constructed and the order of items in a Tuple? > > > > E.g. > > > > FOREACH foo GENERATE $1, $2, $3, $4 > > > > Would that guarantee (or not) that the ordering inside the tuple would > also be ($1, $2, $3, $4) > > > > -RJK > > > > P.S: Im new to Pig so Im probably missing many things. > > > > On 7/22/10 11:56 PM, "Renato Marroquín Mogrovejo" < > [email protected]> wrote: > > > > Thanks there Dmitriy. I will write my own then. > > > > Renato M. > > > > 2010/7/21 Dmitriy Ryaboy <[email protected]> > > > >> that has to be a UDF, there is nothing built in for this. > >> > >> On Wed, Jul 21, 2010 at 6:33 PM, Renato Marroquín Mogrovejo < > >> [email protected]> wrote: > >> > >> > Hey everybody, Does any body know how I can sort a tuple's content? > >> > For example, I have (770001,880001,990001,770001) and I would like to > >> > obtain > >> > (770001,770001,880001,990001). I tried doing a group by the first > field > >> but > >> > the thing is that I still get the whole tuple as a resultant bag. > >> > Thanks in advanced. > >> > > >> > Renato M. > >> > > >> > > > > > > > > -- > Harsh J > www.harshj.com > >
