Well, I guess the good news is we already support 🇯🇵🇰🇷 😬

I think this is the first example we've encountered where a route network has location-specific shields.

It's feeling more and more like we need a route_shield_variant tag in addition to network

One possible solution would be to convince mappers in each country to create a child relation with a country-specific network value like VN:AH. (It wouldn’t be uniform: CN:AH is already taken for Anhui’s provincial routes.) But that’s a lot of national communities to convince, and for something that in their eyes would be easily solvable by data consumers.

Mapbox Streets solves this problem by adding an ISO 3166-1 country code to every feature in its road layer, and sometimes an ISO 3166-2 region code too. With this computed is_in-style information, we don’t need mappers to manually maintain presentational tags on route relations.

There are plenty of other examples of road networks that have different shields across borders, but so far we’ve been getting lucky. For example, the U.S. Route shield looks different in California than it does in other states, but it doesn’t matter for Americana because we always omit legends like “US” and the black background. Over time, we’ll have more reason to distinguish shields by geography:

• The Trans-Canada Highway’s shield is rectangular in British Columbia and Québec but rounded at the bottom in Newfoundland and Labrador.
• Each of Australia’s states has its own shield designs and numbering scheme for national routes.
• The U.S. Bicycle Route System has two different shields. The green one is being used in 15 states on an interim basis until it becomes the national standard in the next MUTCD revision.

I have been looking into this, and unfortunately it seems like the Asian Highway network is largely unsigned in quite a few countries it goes through. I don't know if that inhibits the usability of the shields since the network is supposed to connect between countries, I've had a hard time finding resources which document how useful the network is for wayfinding

So far, which countries leave it largely unsignposted according to your research? It is possible to omit the shield for an arbitrary stretch of highway where a given route is unsigned, by tagging unsigned_ref instead of ref on a subrelation.

It's very hard to tell in the absence of very recent imagery considering the pace of highway development.

I also suspected it's the case that "Asian Highway AH##" is exactly how it's said in some countries, so I don't know if that complicates another comment I'd seen about whether AH is or isn't a part of the ref tag.

Searching social media for the "Asian Highway AH4" compared to other configurations of "route AH4," it's clear that "Asian Highway AH4" is the most common way to refer to that route. (I still can't figure out for sure if and where AH4 is signed, it's well known enough for people to mention in conversations, but there's also not many pictures? I suspect if the answer to this is anywhere it's on YouTube in a long vaguely labeled video and nowhere else.) That makes sense considering a number of countries in the Asian Highway network are former British colonies, so saying "Asian Highway AH4" sounds more correct even though in other countries that might sound incorrect.

I also suspected it's the case that "Asian Highway AH##" is exactly how it's said in some countries, so I don't know if that complicates another comment I'd seen about whether AH is or isn't a part of the ref tag.

The original post above mentions that “AH” or “A” is included in the shield in some countries but not others, so things would already have to vary by country, not to mention the very different-looking shields in each country. I don’t think a compromise shield would work very well. The current network=AsianHighway tagging isn’t precise enough to vary the appearance by county; it would have to be something like network=AsianHighway:XY or network=XY:AH in subrelations, where XY is the country code. Subrelations would also be the way to indicate that part of a route is unsigned and another part is signed.

Searching social media for the "Asian Highway AH4" compared to other configurations of "route AH4," it's clear that "Asian Highway AH4" is the most common way to refer to that route.

Wouldn't you want to search "Asian Highway 4" and/or "Asian Highway route 4" as the null-case for the route prefix in ref? Or is this just the network name?

Right, that all makes sense.

I'll try asking around if I can't find any context for it online. In theory, people who have been on an Asian Highway route recently are at most one person removed away on my contact list, in practice auntie WhatsApp chat has priorities that don't involve highway shields.

The Asian Highway is overlaid on national routes that would be much more well-known to locals in each country, regardless of signage. Vietnam has pretty good AH signage, but it’s still mostly a curiosity to people there, versus the QL and CT numbers (#190).

Searching social media for the "Asian Highway AH4" compared to other configurations of "route AH4," it's clear that "Asian Highway AH4" is the most common way to refer to that route.

Wouldn't you want to search "Asian Highway 4" and/or "Asian Highway route 4" as the null-case for the route prefix in ref? Or is this just the network name?

Yes I did try variations of this since each of them would come up with more mixed results, that was unclear sorry. Like "Asian Highway 4," "Asian route 4." Unscientific method, but what did come up for those was people in countries that don't have AH4 in it and Asian highway routes other than 4.

A mapper in India has confirmed on Matrix (Telegram) that the shield diagrams on Wikimedia Commons are correct for Asian Highways in India.

OpenMapTiles assigns the pseudo-network omt-ie-motorway to roads tagged highway=motorway within Ireland. Could shields by country be accomplished with a similar approach?

I think @ZeLonewolf mentioned that OpenMapTiles was able to accomplish that feat because the British Isles are islands, making it possible to simplify the geometries greatly. To achieve a similar client-side performance benefit in #749, I manually created a convex hull of the international border crossings very loosely based on OSM data.

Ideally, OpenMapTiles would geocode each road to its country for various other purposes, but that’s a lot to ask for, so upstreaming these crude geometries would be a fine alternative. Where does OpenMapTiles stand on supporting the Asian Highway network following openmaptiles/openmaptiles#1648?