Jann Röder <roederja@ethz.ch> writes:
> Ok I now know that it really seems to do what I expected. But I still
> wonder what it does if I use two functions f() and g() that return a
> different number of rows.
You get the least common multiple of their periods. It's ugly, and the
lack of any very sane way to define the behavior is the main argument
for deprecating SRFs in the targetlist.
regards, tom lane