Dear developers,
With the new year underway we thought this would be a good time to review changes that we have in the pipeline to our API, our GTFS feed, and our real-time information.
API
There are some known shortcomings with the API which we want to address. These include:
- Does not include all information in GTFS (like shapes) or adapt to include new additions to GTFS (like route sort order)
- No unified "schedule + predictions" call; no "predictions except for vehicles arriving at the end of their trip" call
- Peak-period slow-down
- Harder to learn than it needs to be with many similar calls and unnecessary nesting
- Has not fully replaced all of the earlier datafeeds
We're working on plans that will address these issues. Look for a request for input on this in the coming weeks and news in the coming months.
GTFS Changes
We want to provide more data in GTFS, in some cases by taking advantage of parts of the spec we're not currently using, and in other cases by using proposed/experimental additions to the spec or proposing our own.
Commuter rail snow schedules
MBTA and its commuter rail contractor Keolis have reduced-service "severe weather" commuter rail schedules ready to go. We want to have them ready to go in GTFS too. In the event of an afternoon announcement that the next day will have a different service schedule we want to be able to quickly publish an updated GTFS feed that reflects that, and load that data into our own systems.
Expect more about this in the next few days.
Multi_route_trips
Some trips should be presented in association with more than one route. For example, trips on route "62/76" should be shown to customers looking at route 62 information or route 76 information. Train 746 is on the "Franklin Line" route in GTFS, but it's really part of both the Franklin Line schedule and the Fairmount Line schedule. We want to recognize this relationship in GTFS and provide whatever information about it is necessary, which will likely mean a new file we're tentatively calling multi_route_trips.txt.
Expect more about this in the next couple of weeks.
Timepoints
GTFS has support for a "timepoint" field, defined as stop_times where the vehicle will hold to schedule to avoid leaving early. We want to use this field and capture when MBTA service does and doesn't do this, which will include recognizing commuter rail "L-note" stops.
Expect more about this in the next couple of weeks.
Platforms and commuter rail child stops
Part of this was discussed in August. We're interested in using the proposed/experimental platform_code field to identify platforms and shortening the stop_name field. Stop_id 70071 would go from having a stop_name "Kendall/MIT - Inbound" to having a stop_name "Kendall/MIT" and a platform_code "Inbound". Additionally commuter rail stations will be restructured with a parent stop and child platforms, using the same conventions and reflecting actual platforms.
Stations: entrances, elevators, transfers
Our representation of stations in GTFS could be richer in a few ways. It could include station entrance locations (existing proposed/experimental element). It could identify the travel time between entrances and platforms, and between platforms and other platforms. Those paths could be identified as accessible or inaccessible (new element). If an accessible path uses elevators those elevators can be identified (new element).
This would allow trip planners to account for all the actual connections to the pedestrian network, and the time required to make transfers. The impact of temporary station changes can be represented within GTFS instead of text messages, and when an elevator fails trip planners can give customers their best options for routing around it.
Stations: bike storage, parking lots, and more
Following on the addition of station entrances, elevators, and transfers, an even richer representation of stations would include bike racks and bike cages, parking lots, kiss-n-ride drop-off locations, bus stop benches and shelters, and more. These would be new elements.
Prediction improvements
Commuter rail arrival/departure time adjustment
MBTA commuter rail trains hold for schedule and do not leave early, unless otherwise noted (L-note.) We want to adjust the commuter rail predictions in MBTA-realtime so that the predicted time is only earlier than the scheduled time if the train is expected to depart early (L-note.)
Commuter rail track assignments and departure times
Commuter rail track assignment at North Station and South Station is the one piece of information available in an older feed but not in MBTA-realtime. We want to remedy that and extend it to Back Bay and, for cases where an inbound train is boarding on an outbound track or vice-versa, all other stations. At the same time we want to incorporate certain kinds of delay data at North Station and South Station directly into predictions.
Mattapan Trolley predictions
Work on Mattapan Trolley real-time information is underway.
Subway prediction improvements
We're working on improvements to predictions near the beginning of the line (adding it to the Green Line and better accuracy on the Red Orange and Blue lines), matching the Green Line trips to scheduled GTFS trips, and improved shuttle handling.
700-series bus route predictions
The MBTA is exploring adding predictions to some of its 700-series bus routes, which are operated by private operators under contract to the MBTA.
--
Thanks for reading this description! We'd love to hear your feedback about this list. We're happy to answer questions too although we ask you to bear in mind that there are a lot of unknowns, many of these are ideas in early stages. We'll have more focused discussions on each item as we design them and then again as we get ready to roll them out. Watch for the first such discussions shortly.
Sincerely,
developer@mbta