[Talk-us] Changing

Paul Johnson baloo at ursamundi.org
Mon Jan 1 03:12:52 UTC 2018

On Sun, Dec 31, 2017 at 8:49 PM, OSM Volunteer stevea <
steveaOSM at softworkers.com> wrote:

> Thanks, Paul.  That is exactly the kind of response (mind blowing in its
> comprehensive completeness, although I, for one, on this channel remain in
> listening mode) I
> was hoping for.

Thanks!  Though I incorrectly identified Sam Baldock Freeway as having
Highway number 1; it's actually 1C.  Whoops.  It was 1 before 1W and 1E
were constructed, and 1 being truncated to somewhere around Springfield and
Eugene, with 1C being I 5 north of that after it was built.

I would also like to see a simplification of exactly the sort you
> describe.  Is it truly as easy as "asserting" that we move routes
> exclusively into relations?  All the routes I deal with are relations, why
> are there still "laggards" who do things differently?  (I'm listening, not
> judging).

The "Standard" renderer still uses the ref=* tag on ways to generate
shields on the main website slippymap, annoyingly, and I've long since
given up on tilting on that windmill on getting carto to work out relations.

What about creating TWO relations for TWO routes?  Like, a type=route
> relation with ref=OR 1W and another type=route relation which is ref=99W
> and network=US:OR?  Are we / is somebody relying upon some older-style
> special-case code in archaic renderer(s) which could/should be updated?

Well, in the Oregon case, the "Highway" ref indeed belongs to the way
itself.  A route might cross multiple different highways.  A route will
frequently traverse multiple different highways (Oregon Route 99E traverses
Highways 1E and 1C, for example, going north to south).  A highway might
not even have a route (like state park roads).  Even some major highways
don't have routes, though Wikipedia incorrectly identifies Oregon Route 141
as a state route, when it's actually State Highway 141, with no route.
Sometimes, they're ramps, and TIGER was voracious importing these.  For
example, http://www.openstreetmap.org/way/70202331 .  It's a flyover ramp,
pretty generic in every regard.  But it is also State Highway 1WR. That's
Oregon for ya...

> I don't want to step on toes (of implementors of old renderers), but
> elegant syntax is elegant syntax.  Let's streamline towards elegant syntax
> where we can

YES!  Thank you.  I have a bone to pick about lane tagging as it relates to
bicycle lanes specifically on this front as well.  How are HOV, bus and
taxi lanes counted but bicycle lanes not?  Zero sense...  Unlike the ref
issue which is relatively major, I switched a couple years ago to including
bicycle lanes in lanes=* and lanes:forward=* and lanes:backward=* and just
added the appropriate lane access tags as necessary, as is done with HOV,
bus and taxi lanes the world over.  One complex example of that would be
http://www.openstreetmap.org/way/361641534 , though turn:lanes is kind of
hamfisted there; a more accurate turn lane tagging would be...


Ground truth makes no logical sense sometimes, and OSM's unrelenting logic
struggles with these cases sometimes.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-us/attachments/20171231/20d80546/attachment.html>

More information about the Talk-us mailing list