SELECT id
    FROM my_table
    WHERE id IN (
     SELECT id
     FROM my_table
     WHERE criteria_a = 19
     ORDER BY create_when DESC
     LIMIT 1000
  );

This is the pattern I am looking for, but I need the criteria_a to be repeated for every value of criteria_a with the important focus being the LIMIT 1000 for any single value of criteria_a. There is no need to put a total LIMIT on the query, just to limit to the 1000 per criteria_a with the specific ORDER BY at that point. Put another way…

SELECT id
    FROM my_table
    WHERE id IN (
          SELECT id
		 FROM my_table
		 WHERE criteria_a = 19
		 ORDER BY create_when DESC
		 LIMIT 1000
	)
       OR id IN (
	  SELECT id
		 FROM my_table
		 WHERE criteria_a = 20
		 ORDER BY create_when DESC
		 LIMIT 1000
     );

Where I desire 2000 total rows. I could turn this into programming code (even a PostgreSQL FUNCTION) that loops over every value of criteria_a and replaces 19 in the example.

I don’t care of it is a JOIN or an IN, I’m more stuck on how to repeat the inner SELECT with the LIMIT 1000 based on sort and criteria_a. Can I do it without looping and/or UNION? Thank you.

  • iZom
    link
    fedilink
    English
    arrow-up
    6
    ·
    1 year ago

    I think you can use RANK as a window function and specify TOP 1000

    • RoundSparrow@lemmy.mlOP
      link
      fedilink
      arrow-up
      1
      ·
      edit-2
      1 year ago

      Good results with this approach. I hadn’t considered the RANK OVER PARTITION BY criteria_a values and it works like a champ. It moves the ORDER BY into the realm of focus (criteria_a) and performance seems decent enough… and it isn’t difficult to read the short statement.

      SELECT COUNT(ranked_recency.*) AS post_row_count
      FROM
        (
           SELECT id, post_id, community_id, published,
              rank() OVER (
                 PARTITION BY community_id
                 ORDER BY published DESC, id DESC
                 )
           FROM post_aggregates) ranked_recency
      WHERE rank <= 1000
      ;
      

      Gives me the expected results over the 5+ million test rows I ran it against.

      If you could elaborate on your idea of TOP, please do. I’m hoping there might be a way to wall the LIMIT 1000 into the inner query and not have the outer query need to WHERE filter rank on so many results?

      • iZom
        link
        fedilink
        English
        arrow-up
        3
        ·
        1 year ago

        Glad this is working for you. Using TOP probably was a bad idea and I think the way you used RANK <=1000 is a better approach.

        If there was a way to safely exclude any of the records - like if you knew that when published was older than X days/months/years it would never make it into the final results, you could filter them out before ranking them. That might squeeze a little more performance out of the query, but could be risky if the data isn’t predictable enough.