Is this a good candidate for plpgsql? - postgresql

Question

Several of the introductory tutorials I've read on using for loops in plpgsql have said unless I really need a for loop, I should find another way.
As a programming novice I can't figure out a better way than a for loop.
I would like to go through each parcel gid, calculate the nearest street intersection by bounding box, then assign the intersection id to the nearest_intersection column in parcels.
All the parts of this function work nicely. But I can't figure out how to put them together.
BEGIN
FOR gid IN SELECT gid FROM parcels
LOOP
UPDATE parcels SET nearest_intersection =
(select intersection.osm_id
from intersections
order by
intersections.geom <-> (select geom from parcels where gid = parcel_id)
limit 1;)
end loop;
end;
Thank you!

Answer 1

In your current code, the loop doesn't make sense indeed because the UPDATE alone already processes every row of the parcels table.
What you probably want is (without a loop):
UPDATE parcels R SET nearest_intersection =
(select intersection.osm_id
from intersections
order by intersections.geom <-> R.geom
limit 1);
which in procedural thinking would be the equivalent to:
for every row R of parcels, find the row in intersections
whose geom is the nearest to R.geom and copies its osm_id into
R.nearest_intersection
On the other hand, if it had to be done with a loop, it would look like this:
FOR var_gid IN SELECT gid FROM parcels
LOOP
UPDATE parcels SET nearest_intersection =
(select intersection.osm_id
from intersections
order by
intersections.geom <-> parcels.geom)
limit 1)
WHERE parcels.gid=var_gid;
end loop;

Answer 2

Don't be a hostage of SQL purism. Write functions with loops. When you are a postgres expert you'll change them to queries. Or not.
You probably missed WHERE clause for UPDATE.