[Talk-us] [Talk-us-newyork] Highway classification guidelines for New York State

[Minh Nguyen]

Vào lúc 03:59 2021-09-13, Eric Patrick đã viết:
>     As far as I can tell, no mainstream OSM-based router directly penalizes
>     a road based on its highway=* value per se. If it lacks a maxspeed=* tag
>     and real-time or historical traffic data is unavailable or unsupported,
>     then the router would assume a speed limit based on the highway=* value.
>       From what I've seen, these assumptions are usually wrong for... just
>     about everywhere.
> 
> 
> There's no direct penalty between the highway levels, it's just how 
> they're assigned as a value. Highway=0, trunk=5, primary=10, etc. The 
> only direct penalty will come when a value=-1. By assigning a value to 
> those roads in that order, a hierarchy is being created as to which road 
> is more important.

Is this a feature request, or perhaps you're familiar with a router that 
places this much importance on the road classification? Mateusz has a 
good point that bicycle routing profiles may place more importance on 
road classification, but if I'm not mistaken, it's a fallback for when 
there aren't more detailed tags like lanes or speed limits.

>       But as elegant as FHWA functional
>     classification may be on its own, shoehorning it into the existing
>     highway=* tagging scheme would not be as clean as using a dedicated key
>     like HFCS=*, because highway=* was originally designed by non-Americans
>     who had no idea about the FHWA's specific functional classifications and
>     it has come to be used by data consumers who also couldn't rely on FHWA
>     definitions.
> 
> IF the HFCS= tag did more than be a label, that would work wonders. I 
> also understand the original designation of highways by the outsiders 
> who didn't know about American highway systems, but I believe we can use 
> what they designed, we can use in different terms. A trunk road will 
> function the same between countries, but the definition may vary. The 
> British have a definition of a trunk road. The road meets this quality, 
> or is assigned as such because of XYZ reason, which works for them. A 
> trunk road definition for America (or any other country for that 
> matter), may be "Primary Arterial roads are trunk roads." The only thing 
> being changed is the definition. The trunk road is out there and is 
> coded as such, we're just looking to change the definition between 
> different countries.

The data model doesn't distinguish between tags used for labeling versus 
tags used for other purposes. In principle, if you prefer roads to be 
rendered based on the functional class without caveats, then you could 
customize one of the off-the-shelf renderers to do so. Some extract 
services may make it more convenient to work with highway=* than a 
regionally relevant key like HFCS=*, but it isn't unheard of for a 
regional renderer to consider very obscure keys for a variety of styling 
properties.

Strictly following the state DOT's functional classifications would 
require some due diligence about copyright status. Some mass-importing 
HFCS=* values could be expected to get permission to copy the DOT's 
maps, but we want casual mappers to have a hand in mapping roads, and I 
think this could be too high a bar for a casual mapper to clear when 
choosing a highway=* preset.

-- 
minh at nguyen.cincinnati.oh.us