tweak benchmark setup ⚡️

- Improve the UX by un-quoting the individual *.sql files and moving the `bench()` call to index.sql.
- Include the *.sql filename in the output CSV.
- Make psql stop on the first error.

benchmark/arrs_deps_by_non_existent_stop.sql

@@ -1,5 +1,3 @@
-SELECT * from bench(
-'SELECT count(*)
+SELECT count(*)
 FROM arrivals_departures
-WHERE stop_id = ''definitely-non-existent'''
-);
+WHERE stop_id = 'definitely-non-existent'

benchmark/arrs_deps_by_route_name_and_time.sql

@@ -1,8 +1,6 @@
-SELECT * from bench(
-'SELECT *
+SELECT *
 FROM arrivals_departures
-WHERE route_short_name = ''S1''
-AND t_departure >= ''2022-08-09T07:10+02'' AND t_departure <= ''2022-08-09T07:30+02''
-AND date >= dates_filter_min(''2022-08-09T07:10+02'')
-AND date <= dates_filter_max(''2022-08-09T07:30+02'')'
-);
+WHERE route_short_name = 'S1'
+AND t_departure >= '2022-08-09T07:10+02' AND t_departure <= '2022-08-09T07:30+02'
+AND date >= dates_filter_min('2022-08-09T07:10+02')
+AND date <= dates_filter_max('2022-08-09T07:30+02')

benchmark/arrs_deps_by_station_and_time.sql

@@ -1,9 +1,6 @@
-SELECT * from bench(
-'SELECT *
+SELECT *
 FROM arrivals_departures
-WHERE station_id = ''de:11000:900100001'' -- S+U Friedrichstr. (Berlin)
-AND t_departure >= ''2022-08-09T07:10+02'' AND t_departure <= ''2022-08-09T07:30+02''
-AND date >= dates_filter_min(''2022-08-09T07:10+02'')
-AND date <= dates_filter_max(''2022-08-09T07:30+02'')',
-40
-);
+WHERE station_id = 'de:11000:900100001' -- S+U Friedrichstr. (Berlin)
+AND t_departure >= '2022-08-09T07:10+02' AND t_departure <= '2022-08-09T07:30+02'
+AND date >= dates_filter_min('2022-08-09T07:10+02')
+AND date <= dates_filter_max('2022-08-09T07:30+02')

benchmark/arrs_deps_by_station_and_time_seq_0.sql

@@ -1,10 +1,7 @@
-SELECT * from bench(
-'SELECT *
+SELECT *
 FROM arrivals_departures
-WHERE station_id = ''de:11000:900100001'' -- S+U Friedrichstr. (Berlin)
-AND t_departure >= ''2022-08-09T07:10+02'' AND t_departure <= ''2022-08-09T07:30+02''
-AND date >= dates_filter_min(''2022-08-09T07:10+02'')
-AND date <= dates_filter_max(''2022-08-09T07:30+02'')
-AND stop_sequence = 0',
-50
-);
+WHERE station_id = 'de:11000:900100001' -- S+U Friedrichstr. (Berlin)
+AND t_departure >= '2022-08-09T07:10+02' AND t_departure <= '2022-08-09T07:30+02'
+AND date >= dates_filter_min('2022-08-09T07:10+02')
+AND date <= dates_filter_max('2022-08-09T07:30+02')
+AND stop_sequence = 0

benchmark/arrs_deps_by_stop.sql

@@ -1,5 +1,3 @@
-SELECT * from bench(
-'SELECT count(*)
+SELECT count(*)
 FROM arrivals_departures
-WHERE stop_id = ''de:11000:900100001::4'' -- S+U Friedrichstr. (Berlin)'
-);
+WHERE stop_id = 'de:11000:900100001::4' -- S+U Friedrichstr. (Berlin)

benchmark/arrs_deps_by_stop_and_time.sql

@@ -1,8 +1,6 @@
-SELECT * from bench(
-'SELECT *
+SELECT *
 FROM arrivals_departures
-WHERE stop_id = ''de:11000:900100001::4'' -- S+U Friedrichstr. (Berlin)
-AND t_departure >= ''2022-08-09T07:10+02'' AND t_departure <= ''2022-08-09T07:30+02''
-AND date >= dates_filter_min(''2022-08-09T07:10+02'')
-AND date <= dates_filter_max(''2022-08-09T07:30+02'')'
-);
+WHERE stop_id = 'de:11000:900100001::4' -- S+U Friedrichstr. (Berlin)
+AND t_departure >= '2022-08-09T07:10+02' AND t_departure <= '2022-08-09T07:30+02'
+AND date >= dates_filter_min('2022-08-09T07:10+02')
+AND date <= dates_filter_max('2022-08-09T07:30+02')

benchmark/arrs_deps_by_time.sql

@@ -1,8 +1,5 @@
-SELECT * from bench(
-'SELECT *
+SELECT *
 FROM arrivals_departures
-WHERE t_departure >= ''2022-08-09T07:10+02'' AND t_departure <= ''2022-08-09T07:30+02''
-AND date >= dates_filter_min(''2022-08-09T07:10+02''::timestamp with time zone)
-AND date <= dates_filter_max(''2022-08-09T07:30+02''::timestamp with time zone)',
-10
-);
+WHERE t_departure >= '2022-08-09T07:10+02' AND t_departure <= '2022-08-09T07:30+02'
+AND date >= dates_filter_min('2022-08-09T07:10+02'::timestamp with time zone)
+AND date <= dates_filter_max('2022-08-09T07:30+02'::timestamp with time zone)

benchmark/arrs_deps_by_time_manual.sql

@@ -1,8 +1,5 @@
-SELECT * from bench(
-'SELECT *
+SELECT *
 FROM arrivals_departures
-WHERE t_departure >= ''2022-08-09T07:10+02'' AND t_departure <= ''2022-08-09T07:30+02''
-AND date >= ''2022-08-08''
-AND date <= ''2022-08-09''',
-10
-);
+WHERE t_departure >= '2022-08-09T07:10+02' AND t_departure <= '2022-08-09T07:30+02'
+AND date >= '2022-08-08'
+AND date <= '2022-08-09'

benchmark/arrs_deps_by_trip_and_date.sql

@@ -1,6 +1,4 @@
-SELECT * from bench(
-'SELECT *
+SELECT *
 FROM arrivals_departures
-WHERE trip_id = ''168977951''
-AND date > ''2022-08-08'' AND date <= ''2022-08-09'''
-);
+WHERE trip_id = '168977951'
+AND date > '2022-08-08' AND date <= '2022-08-09'

benchmark/as-md.js

@@ -15,7 +15,7 @@ pipeline(
 			if (firstRow) {
 				firstRow = false
 
-				const keys = Object.keys(row)
+				const keys = Object.keys(row).filter(key => key !== 'filename')
 				process.stdout.write(`| ${keys.join(' | ')} |\n`)
 				process.stdout.write(`| ${keys.map(_ => '-').join(' | ')} |\n`)
 			}

benchmark/connections_by_non_existent_stop.sql

@@ -1,5 +1,3 @@
-SELECT * from bench(
-'SELECT count(*)
+SELECT count(*)
 FROM connections
-WHERE from_stop_id = ''definitely-non-existent'''
-);
+WHERE from_stop_id = 'definitely-non-existent'

benchmark/connections_by_route_name_and_time.sql

@@ -1,8 +1,6 @@
-SELECT * from bench(
-'SELECT *
+SELECT *
 FROM connections
-WHERE route_short_name = ''S1''
-AND t_departure >= ''2022-08-09T07:10+02'' AND t_departure <= ''2022-08-09T07:30+02''
-AND date >= dates_filter_min(''2022-08-09T07:10+02'')
-AND date <= dates_filter_max(''2022-08-09T07:30+02'')'
-);
+WHERE route_short_name = 'S1'
+AND t_departure >= '2022-08-09T07:10+02' AND t_departure <= '2022-08-09T07:30+02'
+AND date >= dates_filter_min('2022-08-09T07:10+02')
+AND date <= dates_filter_max('2022-08-09T07:30+02')

benchmark/connections_by_station_and_time.sql

@@ -1,9 +1,6 @@
-SELECT * from bench(
-'SELECT *
+SELECT *
 FROM connections
-WHERE from_station_id = ''de:11000:900100001'' -- S+U Friedrichstr. (Berlin)
-AND t_departure >= ''2022-08-09T07:10+02'' AND t_departure <= ''2022-08-09T07:30+02''
-AND date >= dates_filter_min(''2022-08-09T07:10+02'')
-AND date <= dates_filter_max(''2022-08-09T07:30+02'')',
-40
-);
+WHERE from_station_id = 'de:11000:900100001' -- S+U Friedrichstr. (Berlin)
+AND t_departure >= '2022-08-09T07:10+02' AND t_departure <= '2022-08-09T07:30+02'
+AND date >= dates_filter_min('2022-08-09T07:10+02')
+AND date <= dates_filter_max('2022-08-09T07:30+02')

benchmark/connections_by_station_and_time_seq_0.sql

@@ -1,10 +1,7 @@
-SELECT * from bench(
-'SELECT *
+SELECT *
 FROM connections
-WHERE from_station_id = ''de:11000:900100001'' -- S+U Friedrichstr. (Berlin)
-AND t_departure >= ''2022-08-09T07:10+02'' AND t_departure <= ''2022-08-09T07:30+02''
-AND date >= dates_filter_min(''2022-08-09T07:10+02'')
-AND date <= dates_filter_max(''2022-08-09T07:30+02'')
-AND from_stop_sequence = 0',
-50
-);
+WHERE from_station_id = 'de:11000:900100001' -- S+U Friedrichstr. (Berlin)
+AND t_departure >= '2022-08-09T07:10+02' AND t_departure <= '2022-08-09T07:30+02'
+AND date >= dates_filter_min('2022-08-09T07:10+02')
+AND date <= dates_filter_max('2022-08-09T07:30+02')
+AND from_stop_sequence = 0

benchmark/connections_by_stop.sql

@@ -1,5 +1,3 @@
-SELECT * from bench(
-'SELECT count(*)
+SELECT count(*)
 FROM connections
-WHERE from_stop_id = ''de:11000:900100001::4'' -- S+U Friedrichstr. (Berlin)'
-);
+WHERE from_stop_id = 'de:11000:900100001::4' -- S+U Friedrichstr. (Berlin)

benchmark/connections_by_stop_and_time.sql

@@ -1,8 +1,6 @@
-SELECT * from bench(
-'SELECT *
+SELECT *
 FROM connections
-WHERE from_stop_id = ''de:11000:900100001::4'' -- S+U Friedrichstr. (Berlin)
-AND t_departure >= ''2022-08-09T07:10+02'' AND t_departure <= ''2022-08-09T07:30+02''
-AND date >= dates_filter_min(''2022-08-09T07:10+02'')
-AND date <= dates_filter_max(''2022-08-09T07:30+02'')'
-);
+WHERE from_stop_id = 'de:11000:900100001::4' -- S+U Friedrichstr. (Berlin)
+AND t_departure >= '2022-08-09T07:10+02' AND t_departure <= '2022-08-09T07:30+02'
+AND date >= dates_filter_min('2022-08-09T07:10+02')
+AND date <= dates_filter_max('2022-08-09T07:30+02')

benchmark/connections_by_time.sql

@@ -1,10 +1,7 @@
-SELECT * from bench(
-'SELECT *
+SELECT *
 FROM connections
-WHERE t_departure >= ''2022-08-09T07:10+02'' AND t_departure <= ''2022-08-09T07:30+02''
-AND date >= dates_filter_min(''2022-08-09T07:10+02''::timestamp with time zone)
-AND date <= dates_filter_max(''2022-08-09T07:30+02''::timestamp with time zone)
+WHERE t_departure >= '2022-08-09T07:10+02' AND t_departure <= '2022-08-09T07:30+02'
+AND date >= dates_filter_min('2022-08-09T07:10+02'::timestamp with time zone)
+AND date <= dates_filter_max('2022-08-09T07:30+02'::timestamp with time zone)
 ORDER BY t_departure
-LIMIT 100',
-7
-);
+LIMIT 100

benchmark/connections_by_time_manual.sql

@@ -1,10 +1,7 @@
-SELECT * from bench(
-'SELECT *
+SELECT *
 FROM connections
-WHERE t_departure >= ''2022-08-09T07:10+02'' AND t_departure <= ''2022-08-09T07:30+02''
-AND date >= ''2022-08-08''
-AND date <= ''2022-08-09''
+WHERE t_departure >= '2022-08-09T07:10+02' AND t_departure <= '2022-08-09T07:30+02'
+AND date >= '2022-08-08'
+AND date <= '2022-08-09'
 ORDER BY t_departure
-LIMIT 100',
-7
-);
+LIMIT 100

benchmark/connections_by_trip_and_date.sql

@@ -1,6 +1,4 @@
-SELECT * from bench(
-'SELECT *
+SELECT *
 FROM connections
-WHERE trip_id = ''168977951''
-AND date > ''2022-08-08'' AND date <= ''2022-08-09'''
-);
+WHERE trip_id = '168977951'
+AND date > '2022-08-08' AND date <= '2022-08-09'

benchmark/index.sql

@@ -1,5 +1,6 @@
 BEGIN;
 CREATE TEMP TABLE _benchmark (
+	filename TEXT,
 	query TEXT,
 	avg FLOAT,
 	min FLOAT,
@@ -14,7 +15,7 @@ CREATE TEMP TABLE _benchmark (
 
 -- slightly modified from "How to benchmark PostgreSQL queries well"
 -- https://www.tangramvision.com/blog/how-to-benchmark-postgresql-queries-well#sql-function-with-clock_timestamp
-CREATE OR REPLACE FUNCTION bench(_query TEXT, _iterations INTEGER = 100)
+CREATE OR REPLACE FUNCTION bench(_filename TEXT, _query TEXT, _iterations INTEGER = 100)
 RETURNS void
 AS $$
 DECLARE
@@ -41,6 +42,7 @@ BEGIN
 
 	INSERT INTO _benchmark
 	SELECT
+		_filename,
 		_query,
 		round(avg(elapsed)::numeric, 0),
 		min(elapsed),
@@ -58,26 +60,46 @@ END
 $$
 LANGUAGE plpgsql;
 
-\i stops_by_distance.sql
-\i arrs_deps_by_route_name_and_time.sql
-\i arrs_deps_by_station_and_time.sql
-\i arrs_deps_by_station_and_time_seq_0.sql
-\i arrs_deps_by_stop_and_time.sql
-\i arrs_deps_by_trip_and_date.sql
-\i arrs_deps_by_stop.sql
-\i arrs_deps_by_non_existent_stop.sql
-\i arrs_deps_by_time.sql
-\i arrs_deps_by_time_manual.sql
-\i connections_by_route_name_and_time.sql
-\i connections_by_station_and_time.sql
-\i connections_by_station_and_time_seq_0.sql
-\i connections_by_stop_and_time.sql
-\i connections_by_trip_and_date.sql
-\i connections_by_stop.sql
-\i connections_by_non_existent_stop.sql
-\i connections_by_time.sql
-\i connections_by_time_manual.sql
-\i stats_by_route_date.sql
+\set query `cat arrs_deps_by_non_existent_stop.sql`
+SELECT bench('arrs_deps_by_non_existent_stop.sql', :'query');
+\set query `cat arrs_deps_by_route_name_and_time.sql`
+SELECT bench('arrs_deps_by_route_name_and_time.sql', :'query');
+\set query `cat arrs_deps_by_station_and_time.sql`
+SELECT bench('arrs_deps_by_station_and_time.sql', :'query', 40);
+\set query `cat arrs_deps_by_station_and_time_seq_0.sql`
+SELECT bench('arrs_deps_by_station_and_time_seq_0.sql', :'query', 50);
+\set query `cat arrs_deps_by_stop.sql`
+SELECT bench('arrs_deps_by_stop.sql', :'query');
+\set query `cat arrs_deps_by_stop_and_time.sql`
+SELECT bench('arrs_deps_by_stop_and_time.sql', :'query');
+\set query `cat arrs_deps_by_time.sql`
+SELECT bench('arrs_deps_by_time.sql', :'query', 10);
+\set query `cat arrs_deps_by_time_manual.sql`
+SELECT bench('arrs_deps_by_time_manual.sql', :'query', 10);
+\set query `cat arrs_deps_by_trip_and_date.sql`
+SELECT bench('arrs_deps_by_trip_and_date.sql', :'query');
+\set query `cat connections_by_non_existent_stop.sql`
+SELECT bench('connections_by_non_existent_stop.sql', :'query');
+\set query `cat connections_by_route_name_and_time.sql`
+SELECT bench('connections_by_route_name_and_time.sql', :'query');
+\set query `cat connections_by_station_and_time.sql`
+SELECT bench('connections_by_station_and_time.sql', :'query', 40);
+\set query `cat connections_by_station_and_time_seq_0.sql`
+SELECT bench('connections_by_station_and_time_seq_0.sql', :'query', 50);
+\set query `cat connections_by_stop.sql`
+SELECT bench('connections_by_stop.sql', :'query');
+\set query `cat connections_by_stop_and_time.sql`
+SELECT bench('connections_by_stop_and_time.sql', :'query');
+\set query `cat connections_by_time.sql`
+SELECT bench('connections_by_time.sql', :'query', 7);
+\set query `cat connections_by_time_manual.sql`
+SELECT bench('connections_by_time_manual.sql', :'query', 7);
+\set query `cat connections_by_trip_and_date.sql`
+SELECT bench('connections_by_trip_and_date.sql', :'query');
+\set query `cat stats_by_route_date.sql`
+SELECT bench('stats_by_route_date.sql', :'query', 10);
+\set query `cat stops_by_distance.sql`
+SELECT bench('stops_by_distance.sql', :'query');
 
 SELECT * FROM _benchmark;
 

benchmark/run.sh

@@ -7,4 +7,4 @@ set -x
 
 psql -c 'VACUUM ANALYZE'
 
-psql -q -b --csv -f index.sql
+psql -q -b -v 'ON_ERROR_STOP=1' --csv -f index.sql

benchmark/stats_by_route_date.sql

@@ -1,8 +1,5 @@
-SELECT * from bench(
-'SELECT *
+SELECT *
 FROM stats_by_route_date
-WHERE route_id = ''17452_900'' -- M4
-AND date >= ''2022-08-08'' AND date <= ''2022-08-14''
-AND is_effective = true',
-10
-);
+WHERE route_id = '17452_900' -- M4
+AND date >= '2022-08-08' AND date <= '2022-08-14'
+AND is_effective = true

benchmark/stops_by_distance.sql

@@ -1,6 +1,4 @@
-SELECT * from bench(
-'SELECT *
+SELECT *
 FROM stops
 ORDER BY ST_Distance(stop_loc::geometry, ST_SetSRID(ST_MakePoint(9.7, 50.547), 4326)) ASC
-LIMIT 100'
-);
+LIMIT 100