[Talk-us] Route Tagging Consensus

Phil! Gold phil_g at pobox.com
Mon Oct 25 02:03:09 BST 2010


By far the most discussed topics concerned the tagging of numbered
routes.  The discussion split into a couple of different subtopics.


== How to designate routes ==

First was just how route information should be represented.  Almost
everyone agreed on two things: that route shields should be rendered with
network-appropriate shield shapes, not the textual prefixes we have now;
and route relations are the best source of route information.  (NE2 was
the sole dissenter on the last point; he feels that route relations are
too easily broken.  I counted at least Ian Dees, Peter Budny, Anthony,
Paul Johnson, and myself as explicitly voicing support for the use of
route relations.  Anthony went so far as to advocate removing ref= tags on
the roads themselves, though no one else agreed with that and Alex Mauer
argued with him quite vociferously.)

For the ref= tags on ways, Val Kartchner offered the opinion that the tags
should be unambiguously machine parseable, suggesting something like
"US:UT:SR-67".  NE2 seems okay with "I " and "US " prefixes for
Interstates and US Highways, respectively, but suggested no prefix for
state roads and one of "CR ", "CH ", or "CTH " for county roads.  Several
people (Ian Dees and Richard Welty, at least) wanted separate network= and
ref= tags, where ref= only contains the route number and network= works
like the network= tag on route relations.  I don't think I saw a
description of how to handle ref= and network= tags on roads that are
members of multiple routes.  Finally, Craig Hinners suggests having a
separate tag for each network, so US 10 would be tagged
network:country[US]:unitedStatesHighway=10.  (Ian Dees and Val Kartchner
were very opposed to this suggestion, arguing (broadly) that it makes
things harder to deal with.)

Ian Dees also mentioned that he and Lars Ahlzen have been working on
rendering road shields from route relations with Mapnik.

It seems that most people agree that we should be rendering
network-appropriate shields from route relations, and some people are
actively working on that.  In the meantime, I think we should have some
sort of standard for ref= tags and the rendering we have now (or can
easily implement, in contrast to the 100% solution).  There seem to be a
few different options for a consistent, nationwide standard that will fit
in with current European tag usage:

 * "I ", "US ", two-letter state abbreviation, and something for county
   roads (see next section for counties) as prefixes for the route number
   in the ref= tag.  Multiple routes separated by semicolons.
   e.g. 'I 95', 'US 10', 'UT 67'

 * "I ", "US ", "SR ", and something for counties as prefixes for the
   route number in the ref= tag.  Multiple routes separated by semicolons.
   e.g. 'I 95', 'US 10', 'SR 67'

 * "I ", "US ", and something for counties as prefixes for the route
   number in the ref= tag, where state routes don't have a prefix.
   Multiple routes separated by semicolons.
   e.g. 'I 95', 'US 10', '67'

 * Use the textual reference format generally used in the state that
   contains the road to be tagged, preferably as standardized by state
   governments.  This would be "I-nn" for Interstates, except in Texas,
   which would be "IH nn"; "US nn" for US Highways in most states, and the
   state's preferred prefix for state routes (usually either the
   two-letter postal abbreviation or "SR").

 * Only the route number in the ref= tag, and network information in a
   network= tag.  I'm not sure how multiple routes would be represented.
   e.g "ref=95 network=US:I", "ref=10 network=US:US",
   "ref=67 network=US:UT"

 * Separate tags for each network, with the values being the route
   numbers.  Multiple routes in the same network separated by semicolons.
   e.g. "network:country[US]:interstate=95",
   "network:country[US]:unitedStatesHighway=10",
   "network:country[US]:state[UT]=67"

 * Alternately, as above but with appreviated names.
   e.g. "network:US:I=95", "network:US:US=10", "network:US:UT=67".

Keeping in mind that this is a stopgap approach to give us something until
we have proper shield rendering, how much support (or opposition) is there
for each of the options listed above?


== Hyphens ==

Only Val Kartchner offered an opinion on hyphen use, and that was to use
them on Interstates, since that's how they're otherwise written.

Are their objections to changing the wiki to use hyphens for Interstate
ref= tags while continuing to use spaces for other prefixes (assuming we
don't switch to a separate tagging approach entirely)?


== County Prefixes ==

There are two questions here: what sort of prefixes to use in the
prefix-based ref= tagging schemes above, and how to represent county roads
in route relations.

On the question of what to put in way ref= tags, NE2 suggested "CTH" or
possibly "CR" or "CH".  Alex Mauer liked "CTH" because it can't be
confused with a state abbreviation.  Ian Dees suggested "US:ST:CO", where
"ST" is the two-character state abbreviation.

Since there's no clear consensus here, I'd like to ask people what they
prefer (and/or oppose).  Assuming a prefix-based approach to ref= tags on
ways, what do you think of these options for a prefix for county roads
(and possibly also township roads)?

 * CTH (for County Trunk Highway)
 * CR (County Road)
 * CH (County Highway)
 * CO (County)
 * The county abbreviation.
 * Something else (county name, Ian's suggestion, etc.)

On the question of what to use in route relations, several people (Richard
Welty, Toby Murray, and Peter Budny) liked "US:ST:County", where "ST" is
the two-character state abbreviation and "County" is the full county name,
e.g. "US:NY:Albany".  Ian Dees preferred a format like his suggestion for
the ref= tag with the state abbreviation and "CO".  He suggested an
additional is_in tag to indicate *which* county.  Alex Mauer also thought
that a generic county indicator would be best, but suggested a relation to
indicate which county a route belongs in.

Since there's more agreement around the "US:ST:County" approach, are there
objections to just documenting that in the wiki?

-- 
...computer contrarian of the first order... / http://aperiodic.net/phil/
PGP: 026A27F2  print: D200 5BDB FC4B B24A 9248  9F7A 4322 2D22 026A 27F2
--- --
Your father was right.  We ate the wrong son.
                       -- The Host's mother (Angel, "Through the Looking
                          Glass")
---- --- --



More information about the Talk-us mailing list