At 2014-04-02 20:10:54 -0400, robertmhaas@gmail.com wrote:
>
> I think it might underestimate free space relative to tuples because
> the free space map isn't guaranteed to be completely correct. But I
> guess you knew that already...
Yes, and tuple_len is already a slight overestimate (because it counts
the line pointer array as live tuple space for skipped pages). But for
the speedup it provides, I believe the result is still useful.
I'll mention the potential for free-space inaccuracy in the README.
-- Abhijit