PostgreSQL Joining Between Two Values

Question

I have the following tables and am trying to look up county codes for a list of several hundred thousand cities.

create table counties (
  zip_code_from  cha         


        

Answer 1

Months later, this has cropped its head again, and I decided to test some of my theories.

The original query:

select
  ci.city, ci.zip_code, co.fips_code
from
  cities ci
  join counties co on
    ci.zip_code between co.from_zip_code and co.thru_zip_code

Does in fact implement a cartesian. The query returns 34,000 rows and takes 597 seconds.

If I "pre-explode" the zip code ranges into discrete records:

with exploded_zip as (
  select
    generate_series (
      cast (from_zip_code as int),
      cast (thru_zip_code as int)
    )::text as zip_code,
    *
  from counties
)
select
  ci.city, ci.zip_code, co.fips_code
from
  cities ci
  join exploded_zip co on
    ci.zip_code = co.zip_code

The query returns the exact same rows but finishes in 2.8 seconds.

So it seems the bottom line is that using a between in a join (or any inequality) is a really bad idea.