[Talk-se] Göra klart Europa-, riks & länsvägar?

Hanno Hecker vetinari+osm at ankh-morp.org
Thu Sep 9 05:57:28 BST 2010


On Thu, 09 Sep 2010 00:22:48 +0200
Erik Lundin <erik at lists.lun.nu> wrote:
> I'm not completely sure if you mean that the mentioned roundabout 
> mapping is wrong, but as I was the one who did the last edit, I can 
> explain how I think regarding roundabouts and relations:
> * I use to split up roundabouts if not the complete roundabout belongs 
> to the relation.
> * If the way is oneway (or roundabout, which implies oneway), I do not 
> add forward or backward as roles.
As far as I understand 
http://wiki.openstreetmap.org/wiki/Tag:junction%3Droundabout the full
circle is the roundabout (mapping them as four or more segments in an
extra relation is just a proposal). 

Maybe it's a problem with all relation anlyzers and the description in
the wiki ("Role: forward/backward": 'if a route should only be followed
in one direction for some or all of its length, the "role" can indicate
this for some or all of the constituent ways. "forward" means the route
follows this way only in the direction of the way and "backward" means
the route runs only against the direction of the way.'). In our
example, the direction of the way (oneway parts to and from the
roundabout and the roundabout) are clearly in the direction of the way,
so it should not be needed. 
The route analyzers on the other hand, take the first way mentioned in
the relation and see if it has connecting ways, any of the other ways
given in the relation. At a roundabout like in our case it walks the
way until directly after the junction and then has a problem, because
two directions are possible (back on the link road to the junction and
straight on on the normal road). This is reported as hole / segment.
With forward / backward, the way back to the junction would be
ignored (and just tested when it runs the check the other way, i.e.
two checks are made, one forward and one backward). 
See also
http://www.openstreetmap.org/?lat=51.88888&lon=0.89395&zoom=17&layers=00B0FTF
(the forward /backward example given in the wiki
http://wiki.openstreetmap.org/wiki/Relation:route)

	Hanno

> 2010-08-29 14:02, Hanno Hecker skrev:
> > On Fri, 27 Aug 2010 19:30:50 +0200
> > Peter Kindström<peter2010 at infolagret.se>  wrote:
> >> Här tror jag t.ex. min statussida (kommentarerna) plus kanske en
> >> markering på OpenStreetBugs eller en FIXME vore realistiskt idag. För
> >> "kreti&  pleti" kanske min sida kan vara tillräckligt enkel?
> >> (En baktanke jag har är att skriva en artikel i Caravanbladet och be en
> >> stor del av Sveriges campare om hjälp - de vill ha enkelhet i så fall.)
> >>
> >> Vad tycker ni?
> > I've hacked gary68's relation check ([1], [2]) a bit. It does not
> > report an error if a junction=roundabout is in the route... note: this
> > just includes support for closed roundabouts: For example the
> > roundabout north of node 310316 is divided into four parts, none of the
> > links has a forward/backward role for Riksväg 55 (two problems at
> > once :-)).
> >
> > The result can be seen here:
> > http://ankh-morp.org/osm/sv-routes.htm (output restricted to
> > route=road, because route=* would be 4MB :-))
> >
> > 	Hanno
> >
> > [1]:
> > http://svn.openstreetmap.org/applications/utils/gary68/checkrelation.pl
> > [2]: http://wiki.openstreetmap.org/wiki/Relation_Check
> >
> > _______________________________________________
> > Talk-se mailing list
> > Talk-se at openstreetmap.org
> > http://lists.openstreetmap.org/listinfo/talk-se
> 
> 
> _______________________________________________
> Talk-se mailing list
> Talk-se at openstreetmap.org
> http://lists.openstreetmap.org/listinfo/talk-se



More information about the Talk-se mailing list