[OSM-talk] the ref:color schema

[Greg Troxel]

Iván Sánchez Ortega <ivan at sanchezortega.es> writes:

>> As an example, near me in the US you can be on a road which is both I-95
>> S and Massachusetts state route 128 south.  Then there is an
>> intersection where I-95 splits off.  After the interchange you are on
>> 128S and I-93N.
>
> Hm, sounds a little bit like the "Nudo de la Paz"[1] here in Madrid. You can 
> be either in the M-30, M-11, A-1 or M-607 depending on the *lane*.
>
> [1] http://www.openstreetmap.org/?lat=40.48261&lon=-3.68427&zoom=15

Wow, but here there's no lane notion.  The whole road has multiple
designations and it isn't weird.

> OK, but how do we fix this?
> - Do not use administrative levels for highway=* at all, as this info is 
> somewhere else (network=* or even ref:color=* plus country). But then, what 
> is highway=primary for??

highway=primary means that it's not a motorway or close, so that it
isn't divided and restricted access with a high speed limit.  And it
means that it's a road that people use to go significant distances.

Then "ref=es_foo" means it's a foo-type road in Spain, and the talk-es
community can figure out what foo values are appropriate.  This is about
signs and legal status.  (In US, it's
interstate/ushighway/stateroute/countyroute mostly.)

Around me SR 2 is a very important road.  It's motorway in places, trunk
in places, and primary the rest of the way.  It is the main E/W road in
the northern half of Massachusetts.  So it's primary, even though it's
not a US Highgway.

Trying to make primary/secondary and administrative class always match
seems not to be the OSM way - but I'm still pretty new so I could be
off.

> - Use network=*, then assign highway=* depending on the importance of the 
> network (or motorway for physical motorways), then get the shields through a 
> batch process and a huge table.
>
>> > - That table can be a *real* mess.
>>
>> Sure, that's the best argument for needing to encode colors.  It just
>> seems like if one can encode the administrative level etc. and map to
>> colors that's far better - maybe there needs to be a databsae of the
>> mapping that all renderers and other users can get at, sort of logically
>> part of the database but not a node/way/relation.
>
> Do you really think we can work that network->color table out, and make 
> software that fits into osm2pgsql to derive the shields**?

Sorry, shield is US-centric.  "Road sign characteristics" (to include
shape, color, font, etc).

I think the table can be partially worked out, with the rest dealt with
by tagging by exception on the roads that don't follow the rules.

In the US having tags on interstates that say "use normal interstate
sign" seems best avoided.  If signs are not correlated with
administrative designation, then I suppose each relation needs a "road
sign characteristic" tag.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 193 bytes
Desc: not available
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20090419/dd4cf159/attachment.pgp>