Railways: Draft Manual

[michih]

That said, the "main" section of T2 should have "Hegelstraße bound" in the city field and the "Stadthaus" section of T2 should have "Lankow-Siedlung bound" there, with the latter extended to cover the entire route rather than showing just the split section.

My intention with the draft manual was that each direction should be mapped separately but for the whole route, and the names of the terminal station in each direction used to disambiguate. Though looking at what I wrote that perhaps wasn't clear, so what if we change:

- If a service takes a substantially different route in different directions at some point in the middle (e.g. runs around different sides of a downtown loop depending on direction), the different directions of the service must be mapped separately and should have their names distinguished by the terminal station in each direction

to

- If a service takes a substantially different route in different directions at some point in the middle (e.g. runs around different sides of a downtown loop depending on direction), the different directions of the service must be mapped separately and should have their names distinguished by the terminal station in each direction (e.g. "HogExp (Hogwarts bound)" vs. "HogExp (Kings Cross bound)"). Each mapped direction should cover the entire length of the route from terminal to terminal.

That work?

https://github.com/TravelMapping/RailwayData/pull/13/commits/1c648b8f7a898ef7e8934a886a0da51920c55eaf

I'm gone with Hegelstraße bound and Lankow bound now. I thought about eastbound and westbound but dropped the idea very soon.

But how to call the wpt file? And the Abbrev? I'm gone with Lan for the Lankow bound route now. Thoughts?

Edit: It's even more complicated since we already use "Sch" abbreviation. Hmmm....

Code: [Select]

deusnt;DEU-MV;2;;Sch;Schwerin (Hegelstraße bound);deumv.02sch;
deusnt;DEU-MV;2;;Lan;Schwerin (Lankow bound);deumv.02schlan;

[michih]

Other Drafting guidelines
- Lines should be cut into different routes at each crossing of a region boundary, with points labeled as the names of the two bordering regions separated by a slash, e.g. ENG/FRA

ENG/FRA is never a valid border label (and GBR/FRA is a bit of an odd border). Suggest changing the example to NY/CT.

Our rail routes never use "exit numbers" nor "track numbers" at stations or stops (maybe at divergence point but those are not intended to be used). Thus, I think we should rethink the wp label numbering so that the labels are simpler to be read. si404 testd full names w/o truncating. That's easy to read but I think that it is over the top. I'm trying to find something in-between like truncating to 4 or 5 characters by default. I don't say we have to do it this way though. I only want, that we agree on it soon to avoid having to redraft too many routes. Definitely before we "go live for a broader public".

I think about sticking with 3+ letters for street names but expanding to 5+ letters for cities and city districts. For instance, Frankfurt would be truncated to Frank, Berlin would Berlin, i.e. the main stations would be FrankHbf and BerlinHbf. More examples: DortmHbf, DuisbHbf, HambuHbf, HambuAltona, MunchHbf, MunchPasing, BerlinGesundBrunn

It might be a good idea for long-distance and regional services. Not for metro or tram services though.

There's certainly merit in this. I've stuck with 3 letters (and possibly a 4th disambiguating one - often the 5th letter!) in my conversion from 3-letter codes to 3-letter abbrevs, but I see it in say datacheck* and have to think where that is and I know the routes! eg I got 'SKIP_Ber' as a duplicate label on a route and I didn't guess what either of them (Berkswell and Berkhamstead - the latter being not far from where I live!) was until I put the route in the browser and went and saw the stations zoomed in.

*As I looked at this morning. LONG_UNDERSCORE and VISIBLE_DISTANCE are cluttering it up!

Should I give the 5+ city name idea a try?

[Jim]

We are going to have to remove Long_Underscore and Underscore_Labels from the datacheck for rail systems as the formatting for DIV and SKIP labels means lots of these things will exist legitimately and this isn't an error we should be checking for.

Another option, not sure if it's better, might be to use something besides an underscore in the DIV and SKIP labels.

[si404]

Should I give the 5+ city name idea a try?

Worth having a look at it implemented, to help us decide whether it's worth it.

---

Branch handling

Mostly I've tended to treat branches like highways, where the branch ends when it reaches the main route. However I've noticed that the redone/new American systems that branches continue along the line (and so somethings are done like that).

For example:
Currently the London Underground Central line is split into:

• CenLn - from West Ruislip to Epping (as the longest possible route)
• CenLnEal - from Ealing Broadway to North Acton (the junction station)

[li]CenLnHai - from Leytonstone (the junction station) via Hainault to Woodford (the junction station)[/li][/list]

However the off-peak service pattern is the following 3 services (merging short turns in):

• 9tph West Ruislip - Epping and 3tph short-turning Northolt - Loughton
• 3tph between Hainault and Woodford
• 6tph between Ealing Broadway and Hainault with 3tph Ealing Broadway - Newbury Park and 3tph White City - Hainault short turns

Together with the idea that same-line extensions (cf New York commuter rail diesel islands like the NJ Coast Line and Port Jefferson line) should be merged in with the main route even though they are shuttle services at the end of the line (would the same apply to E-BART?), that gives two routes for the browser: West Ruislip - Epping and Ealing Broadway - Woodford.

So is it

• Map the line like a road with branches ending when arriving on the mainline, or
• Map the line by services as two routes with a long overlap across the middle (on other routes this will escalate - the Northern line goes from 3 files to 6 and everything being at least 2 concurrencies)

[michih]

Map the line by services as two routes with a long overlap across the middle (on other routes this will escalate - the Northern line goes from 3 files to 6 and everything being at least 2 concurrencies)

I'm not sure if I got it right. Is it similar to my Hegelstraße bound and Lankow bound thing in Schwerin from yesterday?

[si404]

Map the line by services as two routes with a long overlap across the middle (on other routes this will escalate - the Northern line goes from 3 files to 6 and everything being at least 2 concurrencies)

I'm not sure if I got it right. Is it similar to my Hegelstraße bound and Lankow bound thing in Schwerin from yesterday?

A little, but effectively you have a line with two branches either side of a core section (and a shuttle at the end of one of them), rather than a one-way loop thing in the middle of a line.

[michih]

We are going to have to remove Long_Underscore and Underscore_Labels from the datacheck for rail systems as the formatting for DIV and SKIP labels means lots of these things will exist legitimately and this isn't an error we should be checking for.

Another option, not sure if it's better, might be to use something besides an underscore in the DIV and SKIP labels.

We could go with DIV-AbcDef, DIV/AbcDef or DIVAbcDef. They are not indicated as error in wpt editor.

[michih]

Map the line by services as two routes with a long overlap across the middle (on other routes this will escalate - the Northern line goes from 3 files to 6 and everything being at least 2 concurrencies)

I'm not sure if I got it right. Is it similar to my Hegelstraße bound and Lankow bound thing in Schwerin from yesterday?

A little, but effectively you have a line with two branches either side of a core section (and a shuttle at the end of one of them), rather than a one-way loop thing in the middle of a line.

I'd keep the Ealing branch as-is.
I'm not sure if I got the Hainault loop right. We might merge the loop into the main route and treat the extension to Epping as a branch?

[Jim]

We are going to have to remove Long_Underscore and Underscore_Labels from the datacheck for rail systems as the formatting for DIV and SKIP labels means lots of these things will exist legitimately and this isn't an error we should be checking for.

I've opened a GitHub Issue related to changes we might want to make to have the site update program enable/disable various datachecks as appropriate for different kinds of data.

https://github.com/TravelMapping/DataProcessing/issues/604

[michih]

Should I give the 5+ city name idea a try?

Worth having a look at it implemented, to help us decide whether it's worth it.

It's crap for metros and S Bahn services which mostly run through one city. My city district proposal really sucks.
I've (mostly) implemented it for HH S Bahn now but not for HH U Bahn (metro).

https://github.com/TravelMapping/RailwayData/pull/15/commits/efabe446010a5470e765f6ddb317785754e931c7
https://tmrail.teresco.org/hb/?u=michih&sys=deuhhs

It's different for long-distance services which usually have one stop per city / town only. I don't have such route in Germany though.

[Duke87]

Together with the idea that same-line extensions (cf New York commuter rail diesel islands like the NJ Coast Line and Port Jefferson line) should be merged in with the main route even though they are shuttle services at the end of the line (would the same apply to E-BART?), that gives two routes for the browser: West Ruislip - Epping and Ealing Broadway - Woodford.

So is it

• Map the line like a road with branches ending when arriving on the mainline, or
• Map the line by services as two routes with a long overlap across the middle (on other routes this will escalate - the Northern line goes from 3 files to 6 and everything being at least 2 concurrencies)

Yeah so my thinking here is since we're mapping services, the mapped route should end where the train terminates. There is no such thing as a service on the NYC Subway that runs only from Ozone Park-Lefferts to Rockaway Blvd, that train runs all the way to Inwood 207th... so I made it do that.

This at least is the situation where you only have branching in one direction. Where you have branching in two directions I don't think every single possible service pattern necessarily needs to be mapped (in some cases that could get to be quite a lot) so long as every branch is included in some way. You'll see, for example, the way I've retooled LIRR, I've generally sorted the outward branches between Atlantic, Penn, Grand Central, and LIC based on which one the majority of trains on that branch go to, with a couple branches having both Grand Central and Penn versions because the number of trains to both is near parity. There are unmapped service patterns I could add, but they'd all be 100% concurrent with already mapped patterns and thus redundant.

Re: diesel islands, I'd keep them separate if they are true islands. Wassaic, Port Washington, and Long Branch all have nonzero direct trains to Manhattan which is why I combined them with the inner, electrified portions of their respective lines. Ronkonkoma-Greenport, you may note, is still mapped separately, because there are no direct trains to Manhattan from Greenport.

[michih]

Also related:

There are two branches in Hamburg. Drafted by Si, redrafted by me yesterday.
The branches start at a station but branch off afterwards. At a DIV label. It looks odd on mapview when enabling "Color by concurrencies". Independent of the current "DIV visualization" issue:

https://tmrail.teresco.org/user/mapview.php?u=michih&sys=deuhhu
https://tmrail.teresco.org/user/mapview.php?u=michih&sys=deuhhs

It's fine for casual users but I don't like it for maintenance purpose. I always assume an error since it would most likely be an error for roads.

I don't like the idea but should we start a branch at a DIV wp?

[si404]

I don't like the idea but should we start a branch at a DIV wp?

Hard no.

It will look less weird with the bringing it through.

[michih]

Should I give the 5+ city name idea a try?

Worth having a look at it implemented, to help us decide whether it's worth it.

It's crap for metros and S Bahn services which mostly run through one city. My city district proposal really sucks.
I've (mostly) implemented it for HH S Bahn now but not for HH U Bahn (metro).

https://github.com/TravelMapping/RailwayData/pull/15/commits/efabe446010a5470e765f6ddb317785754e931c7
https://tmrail.teresco.org/hb/?u=michih&sys=deuhhs

It's different for long-distance services which usually have one stop per city / town only. I don't have such route in Germany though.

deuros (Rostock S-Bahn) is better for a test

https://github.com/TravelMapping/RailwayData/pull/16

Things I'd like to get feedback:
- I've implemented the 5+ label idea for the regional deuros system but not for the urban deurot system, i.e. identical wps have different names, e.g. S1's MarienEhe is T1's MarEhe or S1's RostoHbf is T2's Hbf (= Hauptbahnhof = main station)
- I added the "Rosto" prefix to stops of the deuros system in the city of Rostock which had only street names before
- I used the same city/town name abbreviations for DIV labels
- I've nixed the concurrency of the S and T lines in Evershagen since they are different systems on different tracks

Edit: I've also implemented it for Stuttgart S Bahn system.

[Duke87]

I don't like the idea but should we start a branch at a DIV wp?

Hard no.

Agree with the hard no. Route needs to end where train terminates.

Things I'd like to get feedback:
- I've implemented the 5+ label idea for the regional deuros system but not for the urban deurot system, i.e. identical wps have different names, e.g. S1's MarienEhe is T1's MarEhe or S1's RostoHbf is T2's Hbf (= Hauptbahnhof = main station)

My opinion in this is already on record, I won't harp on it.

- I added the "Rosto" prefix to stops of the deuros system in the city of Rostock which had only street names before

Do the official station names that appear on signs include "Rostock"? If not, this shouldn't be there.

If it's necessary to disambiguate otherwise duplicated names, the correct way to do this is with an underscored suffix: "_Ros"

- I used the same city/town name abbreviations for DIV labels

Not picky about this. Whatever works.

- I've nixed the concurrency of the S and T lines in Evershagen since they are different systems on different tracks

While this isn't currently what's broadly implemented it's... yeah this is probably the correct way. I'm good with eliminating concurrencies between physically incompatible systems that are in the same ROW but have no track connection.