Details
Description
See screen shots below - there seems to be a combinatorial explosion at the top of a N1QL ROLLUP for some reason - is that what we want? One screen-shot shows what the counts per group are for a dataset metadata aggregation - the other screen-shot is showing the top of what happens if you try to do a ROLLUP on the same query. Seems odd??? (I've also added a screen shot to show the contents of the metadata dataset dataset that this is running on.)
Seems like it might be a bug - in which case do we need a fix for CC?