regarding this, in gtfs, the station is usually located 5 meters away from the trip shape. so I think that 5 meters is the best cutoff.
Regarding the position, it really depends on the data set. I have found quite accurate data sets and then I also had cases where only one averaged position for several stop positions was used, while the correct positions are 100m apart from each other per direction. Usually, you should not rely on the gtfs feed as only source but at least use some imagery. I would trust mappers more than gtfs feeds and only move the position if the node was not moved after the last import.
Thank you both for your input. I think I will keep the behavior as-is, it works. The ability for mappers to move stops is too important in OSM so overriding always is off the table. When stops move in GTFS, they must be updated in OSM as well - the whole point of this import is to keep data up-to-date.
The only possible point of contention is the 3m radius for resetting to GTFS position. If I hear a good reason to change it (either change the radius or get rid of the behavior entirely), I’ll consider it, but right now it’s not harming anything.
@idshklein I don’t see any connection between that 5m distance from the shape and the distance we’re talking about here. They don’t need to match.
But we can prepare some things in advance
-
GTFS
- who is the onwner of the data
- what is the license of the data, is it compatible with OSM (CC-BY-4.0 is not per se)
- where can I find the *.zip file
- is that a permament link or will that change with every new version of data
- does the server support “last-modified” header of HTTP (curl -sI )
-
PTNA
- is there a single association/authority which provides public transport service or are there multiple, in which regions (counties, cities, …)
- what is the full / short name of the associations/authorities
- “MVV” for “Münchner Verkehrs- und Tarifverbund” in DE-BY (Bavaria, Germany)
- PTNA read a CSV list from OSM wiki where the existing (in reality) routes (bus, tram, train, …) are listed
- any suggestions where to locate that in OSM Wiki
- what are the (search) areas where the associaons/authorities operate
- PTNA uses Overpass API for the search, providing an area for the search (defined via wikidata of the county/city relations)
Sounds complicated? It isn’t!
I can find most of the details once I know the basic information by visiting their web sites
- character sets other than ASCII make it difficult for me though: Arabic, Cyrillic, Hebrew, Thai, … in ASCII aphabetic order
You might find some of the replies in this recent topic relevant: What is the definition of a public transport network?
In short: yes, the Ministry of Transport is in charge of all public transportation in Israel. From what I can tell, they don’t delegate much. There are a handful of different operators, but all the routes are set by the MoT.
Please see List of GTFS feeds - OpenStreetMap Wiki. It’s permanent, but there’s some SSL issues that I explained in that wiki page.
If I ran curl -sl right, it makes no output. curl -sl --cacert ./mot-gov-il-chain.pem https://gtfs.mot.gov.il/gtfsfiles/israel-public-transportation.zip
It’s owned by the Israeli government. The only license I can find is the terms of use, written in Hebrew. I will translate them in a bit, and answer as many more of your questions as I can.
1 Like
My translation (I am not a lawyer): https://www.gov.il/he/pages/gtfs_general_transit_feed_specifications
תנאי שימוש:
על ידי שימוש של דף זה, גישה למידע בדף או הורדת קובץ GTFS הנך מסכים לתנאי השימוש (להלן “תנאי שימוש”), אשר יכולים להשתנות בכל שלב על ידי משרד התחבורה ללא הודעה מראש. במידה ואינך מסכים לתנאי השימוש, כולל שינויים ועדכונים, אינך מורשה להשתמש בדף או להוריד קובץ GTFS. על ידי הורדה של קובץ GTFS, הינך מאשר כי קראת את תנאי השימוש, וההגבלות המפורטות בהמשך.
בנוסף, ומבלי לגרוע מזכויותיה הניתנים למשרד על פי החוק, המשרד לא יהיה חייב בגין כל נזק ו/ או אבדן ו/ או הוצאה מכל מין וסוג שהם אשר עלולים להיגרם כתוצאה משמוש בדף ו/ או במידע ובנתונים המפורסמים בו.
הגם המשרד עושה כל מאמץ על מנת שהחומר המופיע בדף יהא מעודכן ומדוייק, המשרד לא יהיה אחראי לכל טעות, שגיאה, השמטה או חוסר עדכון של המידע ו/או הנתונים המפורסמים בו. כמו כן המשרד לא יישא בכל נזק ו/או הוצאה שיגרמו עקב מעשים או מחדלים כמפורט לעיל.
המשרד שומר לעצמו את הזכות לשנות בכל שלב את תנאי השימוש.
Terms of use:
By using this page, accessing the information on the page or downloading a GTFS file you agree to the terms of use, which may change at any time by the ministry of transport with no prior notice. In case you do not agree to the terms of use, including changes and updates, you are not permitted to uyse the page or download a GTFS file. By downloading a GTFS file, you agree that you’ve read the terms of use, and the limitations detailed below.
In addition, and without reducing from the ministry’s rights as given by law, the ministry will not be liable for any damages and/or loss and/or expense for any kind or sort which may arise as a consequence of using the page and/or the information and data published in it.
The ministry also makes every effort to make the material in the page up to date and accurate, the ministry will not be responsible for any mistake, error, omission or lack of update of the information and/or the data published in it. In addition the minsitry will not be responsible for any damage and/or expence caused by actions or default as detailed above.
The ministry reserves the right to change the terms of service at any point.
Here I’m not entirely sure, maybe others can give more accurate answer. To my understanding, it’s the National Public Transport Authority (הרשות הארצית לתחבורה ציבורית), which is (I think) a subsidiary of the Ministry of Transport. I normally just shorten Ministry of Transport to MoT or MOT.
Can you give an example for other feeds? I don’t know if this exists right now.
For Overpass I use rel(6195356) - Relation: Israel and The Palestinian Territories (6195356) | OpenStreetMap. Ideally, we could use Israel + Judea and Samaria, but a handful of bus stops are in the Palestinian territories of the West Bank (e.g. in Ramallah: Node: מסוף רמאללה (5210692750) | OpenStreetMap), making it Israel + Judea and Samaria + West Bank. The only difference there should be between this combination and the Israel & Palestine relation above is the Gaza Strip.
Make of this mess what you will. I also have no idea whether the MoT has exclusivity over buses in the Palestinian territories of the West Bank, or if the territories have their own bus network.
I forgot to mention: you can hack your way to a “last updated” datetime by parsing the files listing in https://gtfs.mot.gov.il/gtfsfiles/
It updates daily. Do note that it doesn’t specify a timezone. It might be UTC, or IST (Israel Standard Time), or IDT (during DST), or appropriately changing with the seasons. But whichever it is, it’s useful for comparing two versions. My scripts download this index and just compares it with the previous version - if it’s not a perfect binary match, I assume there’s been an update. It would be nice if there’s a cleaner version to do this, let me know.
Yes, the CSV list for PTNA’s source of comparison of the MTA, NYC, NY, USA
Thanks, while using “last-modified” is the recommended “best practice” for GTFS feeds, PTNA knows many other ways to get the YYYY-MM-DD of a source. This one just adds another variant.
For a route relation to be found it is sufficient that at least a single node (lat/lon) is located in the search area: stop, platform, node of way. So starting with as small area will usually be ok.
Providing the relation IDs is very helpful 
Thanks, let’s set it to “IL-MoT” for the GTFS feed and the result of PTNA.
I’m not a lawyer as well, seems to be OK though.
How important is that lowercase o? It’s currently IL-MOT. I’d have to update a bunch of stops to change it to IL-MoT. I’m totally willing to do that, no sweat, but you make the call.
least importance you can think of. I did not start using ‘o’, that was just a suggestion.
1 Like
Then let’s stick with IL-MOT.
1 Like
I don’t understand, where is this CSV supposed to come from? Is there any reason why it would already exist? If not, I think we can assume that it does not exist.
Valid question. Usually, we start with an empty list and fill the list with what PTNA reports as “exists in OSM”. But this can become tedious.
If GTFS data is available, the list can be compiled from that data. Some layout changes here and there (beautifying, headlines, sorting, …) and then the list shows “what exists in reality” (acc. to GTFS though). PTNA then can make an “ACTUAL” versus “TARGET” comparison - what is missing in OSM and which artefacts do we have in OSM versus GTFS.
As an example see the green button on top of the table @ GTFS Analysis for US-NY-MTA-Manhattan-Bus.
The list can then be maintained by (few) local mappers, that’s the idea of the list being stored in OSM wiki. And: this (easier to maintain) list plus PTNA report replaces wiki tables like the one for Utah, USA
Ah, I thought PTNA considers GTFS to be “Actual”. So then, if this list is normally maintained by humans, how does GTFS play into it?
And importantly, for an entire country with thousands of routes, is this approach feasible? Is the file expected to have every route in a region? How does PTNA behave when it doesn’t?
I see your concerns and from my experiences with DE-BY-MVV (a quite huge association) I share them.
PTNA and the CSV list were introduced in 2017. GTFS support in PTNA is quite new, 2-3 years old. So, the master source of information for PTNA is the CSV list, maintained by humans.
Certainly not, especially when the GTFS data changes quite often - we do not consider GTFS changes due to short-term constructions/deviations on road though. How often do GTFS feeds get an update? DE-BY-MVG every day, DE-BY-MVV every 3 months. But even “every 3 months” is hard to maintain for a huge area. I don’t have a solution at the moment. A textual, intelligent comparison of the OSM wiki CSV data with the CSV export of the GTFS could be an approach which will keep the layout of the report.
A direct import of the GTFS data could be a solutioin, but then you still need some way to define the layout of the report: see structure of section 2 of table of Contents of US-NY-MTA
yes, it lists all routes which exist in reality in the area.
They are listed in a separate section (usually section 3) of the report PTNA - US-NY-MTA. But in this section, there is no analysis regarding route_master and members of the route_master: do they match regarding ref, network, operator, colour, …
I think I’m beginning to understand. Israel’s GTFS is updated daily, and I think it can be a master source of information. But I still don’t fully understand— well, to be honest, I think I need a crash course for what PTNA is and how it’s used, in order for this conversation to actually be productive at this point. Got any good intro material somewhere?
I think this is the best bet. If technically possible, perhaps we could split it to different files by district and then by city to make it more navigable. If not, then just laying it out that way would work.
Really hoping for that PTNA crash course 