Hi Michael,
Isn't that also true during the 2nd scan? I'm assuming new inserts during the 2nd scan properly update the index, so couldn't the same mechanism update the index during the 1st scan?
I guess I'm confused because, if you assume a pathological case where all the data gets inserted after the 1st snapshot, the 1st scan wouldn't pick anything up and the 3-step process I had earlier becomes identical to the 2-step process. Is there something special about index_build?
- Josh