Hi!
First of all, you missed the "distinct" in the select. This means that the output dataset would have obs(test) * obs(grid), which probably blows up your space in DATA.
If your datasets are quite large, I recommend to do two steps:
proc sort
data=data.test (keep=id date)
out=data.id_date
nodupkey
;
by id date;
run;
proc sql;
create table data.seconds as
select id, date, seconds
from data.id_date, data.grid
order by id, date, seconds
;
quit;
Now the distinct is not necessary, as that has been dealt with in the first step.
The NOTE about the cartesian joins is expected; it's exactly what we want to do.
... View more
