Nagoya and Shuto urban expressway shields

[1ec5]

Added route shields for expressways, some urban expressway networks, national highways, and prefectural highways in Japan.

Urban expressways, national highways, and prefectural highways in Tokyo.

Expressways, urban expressways, and national highways in Nagoya.

The shield blanks are based on public domain urban expressway, national route, and prefectural route sign blanks on Wikimedia Commons.

[1ec5]

Japan’s urban expressways are toll roads with route naming and numbering reminiscent of bus and subway lines. There appear to be several tagging schemes. This style can only support the following tagging scheme:

• Routes in the Shuto and Nagoya expressway networks are tagged network=首都高速道路 and network=名古屋高速道路, respectively. Though these networks lack a country code prefix, only Japan and Taiwan have route networks suffixed with 高速道路, which means “highway”, and only Japan’s have refs.

This style cannot support the following tagging schemes:

• Routes in the Hanshin expressway network lack network tags but are tagged operator=阪神高速道路 and belong to a collection relation.
• Routes in the Hiroshima expressway network lack network or operator tags but belong to a collection relation.
• Routes in the Fukuoka expressway network lack network or operator tags but belong to route_master relations that are part of a collection relation.

OpenMapTiles doesn’t expose operator. It doesn’t know anything about collection relations, which are frowned upon.

[ZeLonewolf]

The graphics look great. Regarding network=首都高速道路 and network=名古屋高 I'd be concerned about adding support for non-prefixed route networks if we're not supporting them in Europe.

[1ec5]

We’re currently supporting more ambiguous network values from Croatia. At least these networks say exactly which metropolitan area they’re located in.

https://github.com/ZeLonewolf/openstreetmap-americana/blob/a54a346044b36f8a6f3a5706d02bcf9952f0a500/style/js/shield_defs.js#L979-L981

But I agree that these seem to be evidence of a freeform-name approach to network tagging, analogous to public transportation routes as I mentioned above. (The parallels are striking: in Japan, expressway interchanges are conceptualized almost like subway station stops.)

[1ec5]

These Croatian shield definitions were removed in #310.

[1ec5]

There are two tagging schemes for prefectural highways: network=JP:prefectural and network=JP:prefectural:*, qualified by the prefecture name. JP:prefectural is the more common scheme, but both are on the rise.

As far as I can tell, the situation on the ground is similar to county routes in some U.S. states. The shields are uniform nationwide, except for a legend at the bottom that differs by prefecture. However, each prefecture numbers its highways independently, so numbers frequently get reused across borders.

[zekefarwell]

Seems like we (the maintainers) need to make a definitive decision on #149. I'm in favor of taking the approach in #307 over #219 and then dropping support for any other non-prefixed network values we currently support so we are consistent about the standard for inclusion.

[1ec5]

In my opinion, the primary objective should be adequately unambiguous network values. Aesthetic concerns over consistency are purely academic for a key that is only ever read by machines, and given the existing inconsistency even among the various namespaced formats.

From a practical standpoint, the networks in #149 were problematic because one country or other had reserved a generic word or opaque acronym for itself, raising the prospect of a naming collision between unrelated networks. A value like motorway doesn’t even attempt to consider the world beyond the horizon.

No such conflict is conceivable with a network value that spells out a fully qualified name of the network in full in the local language. It’s incredibly unlikely that another city will ever have a reason to name their route network “名古屋高速道路”, which literally means “Nagoya Expressway” in Chinese and Japanese and no other language. In effect, the use of CJK has inherently namespaced this value to a handful of countries, and the inclusion of “Nagoya” geocodes the network more precisely than any code-based namespace in use outside of the U.S.

That said, I understand the discomfort with consuming a network value that looks so exotic because it isn’t in a Latin alphabet. If the prevailing opinion is that a machine-readable key should only be set to Latin values, then the way forward would be to reach out to the Japanese community about choosing new values that conform to this expectation. In the meantime, splitting out the uncontroversial networks in #307 will help focus this discussion.

[ZeLonewolf]

My specific discomfort is in positioning Americana to be the first tool that codifies a particular network scheme, in cases where the network scheme was probably just invented on a whim. If it's shown that a network scheme, though "non-standard" is being consumed elsewhere and has become the norm for historical reasons, I'm okay with saying "yes that's weird but it's an accident of history". But if we're truly the first (as taginfo, and a lack of examples of other users indicate), then I would prefer to wait on those until there's some firm indicator that the tagging scheme in question represents an affirmative position on behalf of the community behind it.