On 12/17/2020 4:05 PM, Alexandre Rafalovitch wrote:
Try with the explicit URP chain too. It may work as well.
Actually in this case we're just making sure uniqueKey is in fact unique in all documents, so default is what we want.
For this particular dataset I may at some future point look into generating ID as a hash of some unique tuple or other, but then I expect we'll still want to keep the UUID fallback.
Dima
