OH: I-76 60 label at south end of Ohio Turnpike Interchange

[Markkos1992]

Should this be 218A to match I-80 or be something else entirely?  I could not see any signs of an Exit 60 on I-76 EB in GSV leading up to this area.

[mapcat]

The interchange there was unnecessarily complicated (due to ramps between the toll plaza and Mahoning Ave). To simplify, I used the examples of Turnpike exit 161 (with ramps to both I-71 & US 42) and Turnpike exit 173 (with ramps to both I-77 and OH 21) and deleted the extra point for Mahoning Ave. The 218 point for both I-76 and I-80 has been moved to the point where the centerlines cross, and 60/218A have been made alt labels for 218.

The Turnpike needs some more work at the other interchanges, most of which lack graph connections, but that's a project for another day.

https://github.com/TravelMapping/HighwayData/pull/2660

[Markkos1992]

I was planning to look at the rest of the Ohio Turnpike within the next few days.  I can type up a list in a separate topic of ones that may need graph connections.

Otherwise, I agree with what you did here.

[yakra]

The lack of graph connections is going to be a natural consequence of the trumpet interchanges, though.

[Jim]

As much as I'd like to fix graph connections, there are many places on toll roads most commonly where it just makes sense to have separate points.

I thought I had a topic about this somewhere here or in a GitHub issue, but I'm not finding it.  The idea in the back of my mind for a few years now is to introduce a new (likely non-clinchable, but eventually preview- or active-level so it would get into graph generation) system to provide the segments that would turn into the missing edges.  The case closest to home for me is NY I-90's 27(NYST) and NY NY30's I-90, which would be connected by an edge that essentially represents the ramp.

A sample ny.nyexit27ramp.wpt (or something like that) file might look like this:

Code: [Select]

NY30 http://www.openstreetmap.org/?lat=42.926807&lon=-74.199686
I-90 http://www.openstreetmap.org/?lat=42.922932&lon=-74.196424


[mapcat]

I was planning to look at the rest of the Ohio Turnpike within the next few days.  I can type up a list in a separate topic of ones that may need graph connections.

Thanks for the offer, but I will do this sometime later this spring.

The lack of graph connections is going to be a natural consequence of the trumpet interchanges, though.

There are a lot of locations where separating the points isn't necessary. Most of these trumpet interchanges connect only two routes.

As much as I'd like to fix graph connections, there are many places on toll roads most commonly where it just makes sense to have separate points.

I thought I had a topic about this somewhere here or in a GitHub issue, but I'm not finding it.  The idea in the back of my mind for a few years now is to introduce a new (likely non-clinchable, but eventually preview- or active-level so it would get into graph generation) system to provide the segments that would turn into the missing edges.  The case closest to home for me is NY I-90's 27(NYST) and NY NY30's I-90, which would be connected by an edge that essentially represents the ramp.

A sample ny.nyexit27ramp.wpt (or something like that) file might look like this:

Code: [Select]

NY30 http://www.openstreetmap.org/?lat=42.926807&lon=-74.199686
I-90 http://www.openstreetmap.org/?lat=42.922932&lon=-74.196424


An additional system of connectors between routes without current graph connections seems like an overly complicated solution to a problem (if this is indeed a problem). If you want graph connections there, why not just put them where the routes cross, rather than introducing extra segments that someone will undoubtedly want to clinch? Introducing another grab-bag system will undoubtedly create more problems down the road.

[Jim]

I am open to all possibilities.

You just hit on some of the reasons I haven't made introducing such a connector system, beyond the fact it would take time to develop.

The only reason it's a problem for me is that if my students are finding the shortest route between two points, it's not going to be possible to use the NY Thruway, much, for example.  I can easily enough explain why that is but it would be nice to remove oddities like this that can cause confusion at first.

[yakra]

http://cmap.m-plex.com/tools/manual_points.php

There are common exceptions to positioning at centerline crossings, such as interchanges where ramps connect nearby, non-intersecting highways, or where a short access road connects a road to another with a trumpet or similar interchange. In these cases, the waypoints for the same interchange on the separate highways cannot be at the same coordinates. Instead, the waypoints should be where the connecting ramps or access road interchange with each highway.

[Markkos1992]

I may think further about this later.  The only interchange on the PA Turnpike offhand that I think could maybe use a graph connection would be New Stanton (Exit 75 due to I-70).

[mapcat]

http://cmap.m-plex.com/tools/manual_points.php

There are common exceptions to positioning at centerline crossings, such as interchanges where ramps connect nearby, non-intersecting highways, or where a short access road connects a road to another with a trumpet or similar interchange. In these cases, the waypoints for the same interchange on the separate highways cannot be at the same coordinates. Instead, the waypoints should be where the connecting ramps or access road interchange with each highway.

So Tim thought this was important. He wasn't using this project for educational purposes. Jim is, and if this will assist him, I'm open to changing the policy. Of course I'd like to hear some arguments in favor of maintaining the current policy first.

[yakra]

The crossroad to receive the point is the connector, not the other numbered route the connector connects to.
Where shall we put ME I-295 28 and ME US1 I-295(28), then?

[michih]

As much as I'd like to fix graph connections

Me too! I hate these gaps! It was hard for me to accept the rule. I remember the discussion with Si....

I thought I had a topic about this somewhere here or in a GitHub issue, but I'm not finding it.  The idea in the back of my mind for a few years now is to introduce a new (likely non-clinchable, but eventually preview- or active-level so it would get into graph generation) system to provide the segments that would turn into the missing edges.

I remember that you and me discussed about the possibility to show the "waypoint for the same interchange with different coordinates" as "Intersecting/Concurrent Routes" link in HB.

I am open to all possibilities.

Well, drafting a new system for trumpets? NO WAY! NEVER!

[dave1693]

I may think further about this later.  The only interchange on the PA Turnpike offhand that I think could maybe use a graph connection would be New Stanton (Exit 75 due to I-70).

Lancaster/Reading/Denver, where the ramp from the Turnpike has an interchange with US222 on its way to its original terminus at what is now PA272 (original US222).

In general, I'm mildly in favor of changing the existing policy in the case of a straight double trumpet interchange. Still thinking about anything more complicated.

[Markkos1992]

I may think further about this later.  The only interchange on the PA Turnpike offhand that I think could maybe use a graph connection would be New Stanton (Exit 75 due to I-70).

Lancaster/Reading/Denver, where the ramp from the Turnpike has an interchange with US222 on its way to its original terminus at what is now PA272 (original US222).

In general, I'm mildly in favor of changing the existing policy in the case of a straight double trumpet interchange. Still thinking about anything more complicated.

I did a quick glance and found quite a few that it may be useful at.  A quick glance at this one tells me no as this road actually becomes the at-grade Colonel Howard Blvd with a diamond interchange at US 222.

[rickmastfan67]

So, we lose mileage on both routes, and an official multiplex on the connector road between the two?  Hmmmm, not a fan of this change to be honest.