Hi Luc,
I've simulated your case and I was able to reproduce it, in 10.1B.
What I noticed:
FOR EACH statement with a join: normal number of reads in VST
Static query with a join: normal number of reads in VST
Dynamic query with a join: the number of reads for the second table in the join is doubled in VST, this number is also the difference between the number of reads in VST for the first table and what the actual figure should be
Without -rereadnolock all three queries give the same result in VST.
Perhaps good to know, the number of index reads for the data is the same for all three queries, with or without the -rereadnolock client parameter.
Strange case...
David