[Enhancement]: Include max/min values in StatsSummary1D?

[diego-hermida]

What type of enhancement is this?

API improvement

What subsystems and features will be improved?

Query executor

What does the enhancement do?

Currently, a StatsSummary1D object includes information to make it easy to compute sum, average and count operations (among others).

For example:

postgres=# SELECT m1 FROM test;
 m1 
----
  1
  2
  3
(3 rows)

postgres=# WITH stats AS (
    SELECT
        stats_agg(m1) AS stats_m1
    FROM test
)
SELECT
    average(rollup(stats_m1)) AS avg_m1,
    sum(rollup(stats_m1)) AS sum_m1,
    num_vals(rollup(stats_m1)) AS count_m1
FROM stats;
 avg_m1 | sum_m1 | count_m1 
--------+--------+----------
      2 |      6 |        3
(1 row)

Would it make sense to include min/max values in StatsSummary1D?

These would be exposed by accessor functions (e.g. max(StatsSummary1D) and min(StatsSummary1D)), allowing users to leverage a single call to stats_agg to compute all the typical aggregation operations: sum, count, avg, min and max.

Thanks,
Diego

Implementation challenges

From what I can see, a StatsSummary1D has a version field, along with n, sx, sx2, sx3 and sx4 that store count, sum, variance/skewness (?) and avg:

postgres=# WITH stats AS (
    SELECT
        stats_agg(m1) AS stats_m1
    FROM test
)
SELECT * FROM stats;
                stats_m1                
----------------------------------------
 (version:1,n:3,sx:6,sx2:2,sx3:0,sx4:2)
(1 row)

Perhaps, adding two more fields sx5 and sx6 to compute the min/max values, respectively? It seems changes are relative to this file: stats1d.rs.

In addition, version field could be used to display a NULL or NaN for version=1 and the actual values if version=2 (for backwards-compatibility).

[Kazmirchuk]

this would be very useful for my project as well