On Mon, Oct 20, 2008 at 11:57 AM, kowsik <[EMAIL PROTECTED]> wrote: > Ok, I thought I understood this, but a little confused now. What > exactly is the effect of passing group=true to a query? I was thinking > it's mainly there to invoke the reduce function, if provided, no? > > In general though, what's the difference between: > > group=true > - when just a map function is provided
Nothing > - when both map and reduce are provided > Values are reduced such that all exactly matching keys are separate Ie, emit(1, 1) emit(1, 2) emit(2, 1) And a reduce that is merely return sum(values) would yield: 1, 3 2, 1 > group_level=1 > - when just a map function is provided Nothing > - when both map and reduce are provided > group_level=1 only works when the emitted keys are arrays, otherwise it behaves as group=true IIRC But, assuming array keys, emit([1, 1], 1) emit([1, 2], 2) emit([2, 0], 0) and again, using return sum(values) as the reduce yields: [1], 3 [2], 0 > Thanks, > > K. > > On Mon, Oct 20, 2008 at 1:08 AM, Paul Carey <[EMAIL PROTECTED]> wrote: >>>> On trunk you can pass ?reduce=false to your query which means you'll >>>> simply get the result of the map. This may be what you want. >> >>> Neat. But, how is this different from omitting group=true in the query? >> >> Simply because if a view contains both a map and reduce function, >> querying that view will by default return the result of the reduce >> function, irrespective of whether group or group_level are absent. >> >> Paul >> >
