A clean rewrite of OpenMBTA.
Assumes PostgreSQL and PostGIS extensions.
gem install pg
To populate the database, download the GTFS data CSV files into data/
and then run db/runall.sh
Then run
psql mbta < trips-today.sql
ssh [email protected] "pg_dump mbta2 -t nextbus_predictions | gzip -c " | gunzip -c | psql mbta2
Include -a if this is repeated
Before optimization of transit_routes on Dec 11 2013:
curl http://openmbta.org/routes/Bus real 0m2.076s user 0m0.006s sys 0m0.004s
After: real 0m0.884s user 0m0.006s sys 0m0.004s
Now to get number 1 trips:
real 0m3.434s user 0m0.010s sys 0m0.021s
[choi@sparta openmbta2]$ time ruby -Ilib lib/transit_trips.rb 'Green Line' 0 > out
real 0m3.406s user 0m2.969s sys 0m0.099s
zoe@li321-67:~/openmbta2$ ruby -Ilib lib/transit_trips.rb 'Green Line' 0 1>/dev/null select stops.stop_name, st.* from stop_times st inner join stops using(stop_id) where trip_id in (select trip_id from trips_today where route_coalesced_name = 'Green Line' and direction_id = 0) order by stop_id, arrival_time, stop_sequence calc_next_arrivals took 1.511933737 seconds make_grid took 3.588673105 seconds fix_grid_stop_ids took 0.000191765 seconds
Still slow:
ruby -Ilib lib/transit_trips.rb 'Green Line' 0 1>/dev/null select stops.stop_name, st.* from stop_times st inner join stops using(stop_id) where trip_id in (select trip_id from trips_today where route_coalesced_name = 'Green Line' and direction_id = 0) order by stop_id, arrival_time, stop_sequence calc_next_arrivals took 2.102558194 seconds make_grid took 3.095658068 seconds fix_grid_stop_ids took 0.000181157 seconds zoe@li321-67:~/openmbta2$ fg