Hi, Speaking about various implementations of some algorithm it reminds me that folks in commons-math faced the same situation and they decided to use Strategy design pattern for this (good choice I think). See http://commons.apache.org/math/index.html#summary item #4.
I am not sure how much this is applicable to M/R environment but I would expect that it should be possible. The only downside of this pattern is that a lot of complex data can be passed between context and strategy even if it is not needed. Other option is to pass context itself into strategy which could be issue with M/R schema depending on the design. Lukas On Fri, Feb 15, 2008 at 8:37 AM, Isabel Drost <[EMAIL PROTECTED]> wrote: > On Friday 15 February 2008, Ted Dunning wrote: > > There should also be an abstract class that adds the desired behavior to > > any data processing algorithm that wants it. > > I think it should even be possible to configure (or inject) the desired > behaviour. The choice of processing does not only depend on the algorithm, > but is also highly dependent on the dataset the user wants to treat. > > Cheers, > Isabel > > -- > Most seminars have a happy ending. Everyone's glad when they're over. > |\ _,,,---,,_ Web: <http://www.isabel-drost.de> > /,`.-'`' -. ;-;;,_ > |,4- ) )-,_..;\ ( `'-' > '---''(_/--' `-'\_) (fL) IM: <xmpp://[EMAIL PROTECTED]> > -- http://blog.lukas-vlcek.com/
