The core of the partitioner is the comparator you return. Everything will still work (for now) if midpoint just throws unsupportedoperation. It's going to be used by the anti-entropy code, though, so implementing a non-toy partitioner that can't be midpointed is a Bad Idea.
-Jonathan On Sat, Oct 10, 2009 at 8:37 AM, Joe Stump <j...@joestump.net> wrote: > As I said in my email, it's a code test. Simply meant to test an applicant's > skills during the hiring process. > I'm aware of OPP and COPP and choosing partitioners. > > --Joe > On Oct 10, 2009, at 7:02, Mark Robson <mar...@gmail.com> wrote: > > > 2009/10/10 Joe Stump <j...@joestump.net> >> >> I've got a guy doing a code test for us and he has some questions about >> custom partitioners: >> http://gist.github.com/205537 >> >> Wondering if anyone could chime in. > > > I'm curious as to why you don't just use the OrderPreservingPartitioner and > apply the transformation to the email addresses in the application before > you use them as keys. > > One of the problems with Cassandra is that you can only use one partitioner > for the whole cluster. > > When I've looked at using it, it has always been clear that key choice is > vital for range scans to be at all useful. > > Mark >