PostgreSQL trigger execute procedure or join on NEW - postgresql

Question

-- trigger function
CREATE OR REPLACE FUNCTION fn_cities_geo_update_event() RETURNS trigger AS $fn_cities_geo_update_event$
BEGIN
is it ok to set a value (geog) which will be used later in the function?
NEW.geog := ST_SetSRID(ST_MakePoint(NEW.longitude,NEW.latitude), 4326)::geography;
this is one way that I tried to find the average of all cities within 90km of a new city so that the new city will have data populated about it
if NEW.rent_avg IS null then
NEW.rent_avg = (
SELECT avg(a.rent_avg)
FROM cities as a
-- I think I'm missing something here... ?
ST_DWithin(a.geog, NEW.geog, 90000)
);
end if;
here is another way that I tried:
if NEW.food_avg IS null then
NEW.food_avg := (
SELECT avg(a.food_avg)
FROM cities AS a
JOIN NEW AS b
ON ST_DWithin(a.geog, b.geog, 90000)
);
end if;
RETURN NEW;
END;
$fn_cities_geo_update_event$ LANGUAGE plpgsql;
but neither worked.
edit: here is a copy of the table that I'm working with
l |u |n |population|unesco|r |c |rent_avg|rent_low|rent_high|food_avg|food_low|food_high|transport_avg|transport_low|transport_high|k |i |quality|hcid |hc |latitude |longitude |spread |density |distance |dbn |state|geog |id
-----|--|--------------------------|----------|------|-------------------------|----------------|--------|--------|---------|--------|--------|---------|-------------|-------------|--------------|---------|---|-------|-------------------------------------|----|-----------|----------|-----------|----------|----------|--------------------------|-----|--------------------------------------------------|-----
false|NZ|Gisborne | 34274| 0|Australia and New Zealand|New Zealand | 92.2430| 51.1720| 143.4150| 22.0300| 13.3190| 35.3490| 7.0650| 5.9800| 13.0450|4VHV8X00+|GIS| 1712|place:Gisborne | 46|-38.6640015|177.977005| 0.99940002| 0| | | |0101000020E6100000000000A0433F664000000000FE5443C0| 1611
true |NZ|Patutahi | 386| |Australia and New Zealand|New Zealand | | | | | | | | | |4VHV9V00+| | 1000|place:Patutahi | 35|-38.6170006|177.899994| | | 8.5|Patutahi | |0101000020E6100000000000C0CC3C6640000000E0F94E43C0| 1624
true |NZ|Waihau Bay | | |Australia and New Zealand|New Zealand | | | | | | | | | |4VJV8Q00+| | 1000|place:Waihau_Bay | 6|-37.6780014|177.796005| | |110.699997|Waihau Bay | |0101000020E6100000000000E078396640000000C0C8D642C0| 1671
true |NZ|Tokomaru Bay | 495| |Australia and New Zealand|New Zealand | | | | | | | | | |4VHWV800+| | 1000|place:Tokomaru_Bay | 5|-38.1329994|178.300003| | |65.4000015|Tokomaru Bay | |0101000020E6100000000000A09949664000000020061143C0| 1673
true |FR|Cornebarrieu | | |Western Europe |France | | | | | | | | | |8FM3M800+| | 1000|place:Cornebarrieu | 112| 43.6559982|1.33299994| 3.60581994| | 3.5999999|Cornebarrieu | |0101000020E6100000000000C0F753F53F000000C0F7D34540| 6070
edit: create trigger statement
DROP TRIGGER IF EXISTS tr_cities_inserted ON cities;
CREATE TRIGGER tr_cities_inserted
BEFORE INSERT ON cities
FOR EACH ROW
EXECUTE PROCEDURE fn_cities_geo_update_event();

Answer 1

Okay, this is interesting.. I am new to triggers (as you may have noticed ^^) so it seems, when I recreate a table AFTER I created a trigger for that table, the trigger doesn't work. It just returns null for the respective columns.
When I stick to the proper order, I get it to work with my sample reproduction of your case:
CREATE TABLE so_postgres
(n text,
r text,
c text,
rent_avg numeric,
food_avg numeric,
transport_avg numeric,
latitude numeric,
longitude numeric,
geog geography
);
INSERT INTO so_postgres
VALUES ('Möhringen', 'central europe', 'germany', 200.45, 56.45, 4.56, 48.725866,
9.146131, ST_SetSRID(ST_Point(9.146131, 48.725866), 4326)),
('Vaihingen', 'central europe', 'germany', 155.33, 44.12, 2.78, 48.732550,
9.108291, ST_SetSRID(ST_Point(9.108291, 48.732550), 4326)),
('Sigmaringen', 'central europe', 'germany', 298.11, 59.67, 1.99, 48.090797,
9.230243, ST_SetSRID(ST_Point(9.230243, 48.090797), 4326));
CREATE OR REPLACE FUNCTION fn_cities_geo_update_event() RETURNS trigger AS $fn_cities_geo_update_event$
BEGIN
NEW.geog := (ST_SetSRID(ST_MakePoint(NEW.longitude,NEW.latitude), 4326)::geography);
if NEW.rent_avg IS null then
NEW.rent_avg := (
SELECT round(avg(a.rent_avg), 2)
FROM so_postgres as a
WHERE ST_DWithin(a.geog, NEW.geog, 50000)
);
end if;
RETURN NEW;
END;
$fn_cities_geo_update_event$ LANGUAGE plpgsql;
CREATE TRIGGER fn_cities_geo_update_event BEFORE INSERT OR UPDATE ON so_postgres FOR EACH ROW EXECUTE PROCEDURE fn_cities_geo_update_event();
INSERT INTO so_postgres (n, r, c, latitude, longitude)
VALUES ('Degerloch', 'central europe', 'germany', 48.725866, 9.146131);
n | r | c | rent_avg | food_avg | transport_avg | latitude | longitude | geog
-------------+----------------+---------+----------+----------+---------------+-----------+-----------+----------------------------------------------------
Möhringen | central europe | germany | 200.45 | 56.45 | 4.56 | 48.725866 | 9.146131 | 0101000020E610000012DDB3AED14A2240A1A3552DE95C4840
Vaihingen | central europe | germany | 155.33 | 44.12 | 2.78 | 48.732550 | 9.108291 | 0101000020E6100000FBE6FEEA71372240A857CA32C45D4840
Sigmaringen | central europe | germany | 298.11 | 59.67 | 1.99 | 48.090797 | 9.230243 | 0101000020E61000000F441669E275224097C9703C9F0B4840
Degerloch | central europe | germany | 177.89 | | | 48.725866 | 9.146131 | 0101000020E610000012DDB3AED14A2240A1A3552DE95C4840
To answer your last comment: I add the ST_DWithin to the rest of the query via a WHERE-clause.
Does it work for you?