[Talk-us] [josm-dev] Relation editor support for north/south and east/west similar to forward/backward

Sat Nov 30 09:10:29 UTC 2013

Martijn

I'm good with having a separate discussion of milepoints/*pointes
kilometriques, *sure.  I'll probably wait a week or two until a consensus
emerges on posted directionality, as you suggest.

Peter

On Thu, Nov 28, 2013 at 10:37 AM, Martijn van Exel <m at rtijn.org> wrote:

> Peter,
> I think we should separate the discussion related to linear
> referencing / mileposts from the cardinal direction discussion - these
> are two different things really, to my mind. The notion of cardinal
> direction is a relatively straightforward one, and that is already
> cause for (cultural) confusion. Introducing the GIS concept of linear
> referencing into this discussion I think adds to the confusion. We
> should perhaps discuss that separately - I for one don't see the
> immediate relation between the two, but I am happy to be proven wrong.
>
> On Wed, Nov 27, 2013 at 3:08 AM, Peter Davies <peter.davies at crc-corp.com>
> wrote:
> > Martijn
> >
> > I, too, await your clarification for KristenK, as I'm a little confused
> too.
> >
> > We need to keep in mind that positive and negative GIS Linear Reference
> > directions (which are handy as global solutions applying everywhere in
> the
> > US at least) beginning at milepoint 0.0, usually on the southern or
> western
> > state boundary for rectangular states, are not the same as posted DOT
> miles
> > that sit on green and white pressed steel signs on the shoulder of all
> > Interstates and many state/US routes. DOT miles often jump and can
> > occasionally change directions, as route designators are altered
> (something
> > that happens quite often) and bypasses are built.  The cost of reporting
> the
> > whole route is usually prohibitive.
> >
> > So GIS LRS positive and (imperfect) posted DOT miles are handy things to
> > keep in mind as long as we realize that there are always a few
> exceptions to
> > break our defaults.  Similarly, posted cardinal directions are fairly
> > rules-bound but certainly not 100%. This is why I think a good OSM
> solution
> > needs to be explicit rather than implicitly inferred from highway
> geometry.
> >
> > Examples of state GIS definitive records are built by ESRI "Roads and
> > highways" (used in Indiana) and by Agile Assets (used in Idaho).  Check
> out
> > http://www.esri.com/software/arcgis/extensions/roads-and-highways
> >
> > Peter
> >
> >
> > On Tue, Nov 26, 2013 at 5:24 PM, Kristen Kam <kristenk at telenav.com>
> wrote:
> >>
> >> Martijn,
> >>
> >> I want to make sure I understand what you're trying to convey to the
> >> group. Are you saying that If a way has a member role value of "east"
> >> then east will mean forward and then west (it's opposite) would mean
> >> backward?
> >>
> >> Example logic:
> >>
> >> ** If member role = east, node direction is eastbound would mean
> >> forward and backward would mean 'west'
> >> ** If member role = west, node direction is westbound would mean
> >> forward and backward would mean 'east'
> >> ** If member role = north, node direction is northbound would mean
> >> forward and backward would mean 'south'
> >> ** If member role = south, node direction is southbound would mean
> >> forward and backward would mean 'north'
> >>
> >> If the logic I stated above successfully captured with your
> >> suggestion, then I would like to expand on it. Why not just make the
> >> cardinal direction value-forward/backward value relationship a bit
> >> more simpler? I would like to cite Peter Davies' discussion on the
> >> Highway Directions in the US wiki page. He stated that milepoints
> >> increase as highways that trend northward or eastward--say positive
> >> direction. So if one is traveling south or west on a highway, the
> >> milepoints are decreasing--say negative direction.
> >>
> >> With this in mind, couldn't we just say that north/east = forward
> >> (forward movement is positive!) and west/south=backward (backward
> >> movement is negative!)? If we're digitizing our edges, the suggestion
> >> would be to set the node direction of two-way, aka single-carriageway
> >> roads, into a positive direction and the member roles values to north
> >> or east. Basically what you did for
> >> http://www.openstreetmap.org/browse/relation/2308411, but setting the
> >> single-carriageway/two-way roads to 'east' instead of 'west'.
> >>
> >> Thoughts Martijn? Others??
> >>
> >> Best,
> >>
> >> Kristen
> >> ---
> >>
> >> OSM Profile → http://www.openstreetmap.org/user/KristenK
> >>
> >>
> >> -----Original Message-----
> >> From: Martijn van Exel [mailto:m at rtijn.org]
> >> Sent: Tuesday, November 26, 2013 2:47 PM
> >> To: Ian Dees
> >> Cc: Florian Lohoff; OpenStreetMap-Josm MailConf; OSM US Talk
> >> Subject: Re: [Talk-us] [josm-dev] Relation editor support for
> >> north/south and east/west similar to forward/backward
> >>
> >> Yes, sorry for not being clearer. As Ian indicates, this is the
> >> *signposted cardinal direction* of a numbered road route, which does
> >> not change with the actual compass direction of the road. The guiding
> >> principle for the United States is that the odd numbered Interstates
> >> are north/south, and the even numbered Interstates are east/west. This
> >> is independent from the local compass direction. So for example, I-80
> >> is east-west, but runs almost north-south locally (for example here:
> >> http://www.openstreetmap.org/browse/way/203317481) but the sign would
> >> still say 'I-80 East' (or West as the case may be).
> >>
> >> So the relation between the east--west and north--south member roles
> >> is equivalent to the relation between forward--backward.
> >>
> >> Because the cardinal direction is commonly included on the road signs
> >> (see example
> >> http://www.aaroads.com/west/new_mexico010/bl-010_eb_at_i-010.jpg)
> >> this information is useful in the U.S. (and Canadian) context as a
> >> drop in replacement for the traditional forward / backward role
> >> members.
> >>
> >> Hope this clarifies somewhat!
> >> Martijn
> >>
> >> On Tue, Nov 26, 2013 at 2:57 PM, Ian Dees <ian.dees at gmail.com> wrote:
> >> > On Tue, Nov 26, 2013 at 3:51 PM, Florian Lohoff <f at zz.de> wrote:
> >> >>
> >> >> On Tue, Nov 26, 2013 at 12:30:25PM -0700, Martijn van Exel wrote:
> >> >> > Hi all,
> >> >> >
> >> >> > I'm new to this list so please bear with me.
> >> >> > The relation editor currently only parses 'forward' and 'backward'
> >> >> > roles when considering the visual representation in the rightmost
> >> >> > column. In the United States, north/south and east/west are very
> >> >> > common as member roles for road routes, because that is how they
> >> >> > are officially signposted.
> >> >>
> >> >> I would be very careful in using this. Is this really "south" e.g.
> >> >> 180° ? Or is it more like 99° ? Or 269° ?
> >> >>
> >> >> Most streets are not strictly on the 90° raster and signposts are
> >> >> only rough directions.
> >> >>
> >> >> Addings this to OSM might make it much more difficult for Data
> >> >> Consumers to process and interpret data.
> >> >
> >> >
> >> > No, these aren't compass directions. They're the directionality of the
> >> > road.
> >> > For example, this way is part of the I-94 interstate going west, but a
> >> > compass in a car driving on it would tell the viewer they were
> >> > pointing
> >> > north:
> >> >
> >> > http://www.openstreetmap.org/browse/way/39372612
> >>
> >>
> >>
> >> --
> >> Martijn van Exel
> >> http://oegeo.wordpress.com/
> >> http://openstreetmap.us/
> >>
> >> _______________________________________________
> >> Talk-us mailing list
> >> Talk-us at openstreetmap.org
> >> https://lists.openstreetmap.org/listinfo/talk-us
> >
> >
>
>
>
> --
> Martijn van Exel
> http://oegeo.wordpress.com/
> http://openstreetmap.us/
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-us/attachments/20131130/1a8e13c7/attachment-0001.html>