> Why make the index unique when the btree allows duplicates? in normal btrees keys are unique. in btrees-with-duplicates key+value pairs are unique.
that's the idea. for normal btrees we create index on keys only, for dup-btrees -- index on pairs. allowing duplicated key+value pairs will totally break the way cursor works, because they assume that key+value pair identifies current row. (and perhaps it will break some other things, like remove-kv-pair). it's possible to implement duplicated pairs support, but that's pretty hard, and i don't see any practical reason for this. > This breaks inverted indices with duplicate values. it does not. value->oid pairs are always unique. _______________________________________________ elephant-devel site list elephant-devel@common-lisp.net http://common-lisp.net/mailman/listinfo/elephant-devel
