Kyotaro HORIGUCHI <horiguchi(dot)kyotaro(at)lab(dot)ntt(dot)co(dot)jp> writes: > Hello, I found that tuples come from file_fdw has strange xmin and xmax.
file_fdw isn't documented to return anything useful for xmin/xmax/etc, so I don't find this surprising.
> The patch attached fixes the header of materialized tuple to be > sane (2, 0) if the source slot was a virtual tuple in mechanism().
I don't really think it's ForeignNext's place to be doing something about this. It seems like a useless expenditure of cycles. Besides, this fails to cover e.g. postgres_fdw, which is likewise unconcerned about what it returns for system columns other than ctid.