[Talk-us] Proposal: Sunset ref=* on ways in, favor of relations

Paul Johnson baloo at ursamundi.org
Wed Nov 11 12:31:55 UTC 2015


On Mon, Nov 9, 2015 at 9:07 AM, Richard Welty <rwelty at averillpark.net>
wrote:
>
> there's a chicken and egg problem here. you need mappers to work to fix
> up route relations but it's hard to get motivated unless something visibly
> happens with them.
>

Agreed.  I'm aggressively trying to ensure completion on relations in
Oklahoma right now as prepwork on this, especially since worst case
scenario, at least having the relations in place would make it scriptable
(or at least JOSM-based semi-automatically) possible to fix the ref=* tags.


> 1) update the stylesheets for the proof-of-concept; they need to be
> based on the current carto sheets going forward (if i remember correctly,
> the proof-of-concept is pre-carto.)
>

To the best of my knowledge, this is correct; a lot of work that was moving
forward was back-burnered for other projects pending carto's update as
well, so now we're just waiting for the folks who understand this to check
the back burners again (or another cook to enter the kitchen).


> 2) bring up a map based on this up on a US chapter website, clearly
> stating that its a demo/proof of concept for US shield rendering
>

Or just a US-style of the whole world.  Granted, the North American
audience tend to take the whole "route shield" thing farther than most,
though as a result, the North American audience is going to be the most
likely to expect this by default regardless where in the world they look.
Case in point, at least the US version of Google Maps is highly consistent
about using shields worldwide, such as the blue hexagons for German
autobahns, the green shields for Japanese highways, maple leaves on the
Trans Canada highways, and arrowheads on Mexican federal highways (though
Google uses generic circles for state highways and rectangles for
county/city highways, which leaves a lot of edge case clashes that I tend
to pick up on since I seem to have a (admittedly fairly annoying) knack for
finding weird edge cases).


> 3) make the needs/requirements for growing out the shield rendering
> well known to the community. issues that require some care would include
>
> a) svg files for shield types that are currently not represented (mostly
> obscure county route signage)
>
> b) the mapping files for shield->route (again, mostly county routes as
> far as i know.)
>

Not quite sure what you mean there in 3b), the mapping of highway type is
in the network=* tag, such as network=US:US for a US highway, or
network=US:OK for a state highway in Oklahoma.


> the result will be that US based mappers will have a place they can go
> look to see the results of their work. this, i think, would be a pretty
> powerful motivator, and it doesn't require any action on the part of
> the maintainers of the map on openstreetmap.org
>
> it might be that the shield based rendering sticks on the US map
> and never makes it to the international map, and that would be ok.
>

Heck, just the search-ability with tools like the XAPI would benefit the
global mapping community, even in regions where the community isn't
concerned about multiplexed routes across modes (such as a way that has a
bus route and cycle route that also has a highway designation).  Though
this does lack the immediate gratification for folks who aren't prone to
ask questions like, "what does the entire national highway system in South
Sudan look like by itself?" or "Where is Missouri supplemental highway KK
located?".  Would also be handy for data consumers like Wikimedia
Foundation's Maphack, which could pull routes from an API mirror for pages
concerning such routes (like, say, the Wikipedia page "Interstate 70 in
Kansas").

Having this information in relations like we already have for other routes
would also eliminate the need to way-tag the same information, much as
relations have already done this from the get go for transit information
and since the introduction of relations first deprecated then obsoleted way
tags like lcn_ref=* and lcn=yes.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-us/attachments/20151111/f33990a0/attachment.html>


More information about the Talk-us mailing list