I'd suggest something like getting a list of all the records, sorting that list by whatever content you need to test in order to recognize redundant entries, then processing that list for uniqueness. I beleve I've seen an example of this kind of approach (building a keyword index, or something like that), but I can't seem to find it right now.... In general, avoid anything that involves convolving one node against many other nodes. That becomes an N-squared problem all too easily.
- Performance problem in transformation?? Shashank Rajvanshi
- Joseph_Kesselman
