[OSM-talk] How do I tag this road in OSM

Etienne 80n80n at gmail.com
Mon Jul 31 22:38:41 BST 2006


On 7/31/06, Christoph Burghardt <Christoph.Burghardt at uni-rostock.de> wrote:
>
> On Mon, 31 Jul 2006 12:40:35 +0100, David Groom wrote:
>
> >
> > ----- Original Message -----
> > From: "Christoph Burghardt" <Christoph.Burghardt at uni-rostock.de>
> > To: <talk at openstreetmap.org>
> > Sent: Monday, July 31, 2006 11:24 AM
> > Subject: Re: [OSM-talk] How do I tag this road in OSM
> >
> >
> >> What is the gain of doing this instead of just tagging the segments
> >> indivually?
> >
> > 1) It gives more flexiblility, as a segment can be shared acrosss
> several
> > ways.
>
> If i don't have any ways i don't need to share a segment. thats would be
> a reason for removing ways!
>
> > 2) Surely it also reduces the amount of data the database has to hold.
> > Lets say there is a street comprised of twenty segments. If you were to
> > tag each segment with say "highway = primary, abutters = residential,
> > name = high street, ref = A3055", then each of those tags has to be
> > stored twenty times.  If however you created one way from the twenty
> > segments, then tagged the way, those tags only get stored once.
>
> that depends on the underlying table layout. on the other hand you would
> save some data, because you don't need the way-segment mapping anymore. if
> you take the example, asked from the thread starter, you have to tag each
> segment different. generating 4 (or even 5 ways) increases your storage
> in this case. but i don't think anyone would care about these bytes.
>
> > 3) If the "high street" segments are tagged 20 times, then if you do a
> > search for high street you get 20 results rather than the more
> > meaningful one result.
>
> a way is just a collection of segments. you can easily filter out the
> duplicate entries. a way is in no way a primary key, bigger town have
> more ways with the same name.
>
> >
> > 4) If you tag segments with names rather than tagging a way, then
> > Osmarender will try and display the name 20 times, which doesn't look
> > good, particularly is segments are short (such as round a bend)
>
> Thats the "fault" of osmarender and can be easily changed.
>
> I don't want to argue for removing ways. But when ways are only a
> fast index of grouped segments, you could automate the task of generating
> this index. And i hate doing something manually, which is just stupid
> machine work.


You might want to think about it the other way round.  Segments are the
redundant element.  All they provide is information about the direction
between two nodes.  Ways can the thought of as multi-segmented paths.

When two ways share a segment there is a problem.  The direction has to be
the same for both ways.  This is often not correct.  Consider a one way
street with a cycle lane in the opposite direction.  If a way was made up of
an ordered list of nodes then this problem would go away.

It is likely, IMHO, that segments will be removed rather than ways.  I'm
contemplating making the next major version of Osmarender ignore segments
completely (including all segment level attributes).

Etienne


Christoph
>
>
> _______________________________________________
> talk mailing list
> talk at openstreetmap.org
> http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/talk
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20060731/ea0be064/attachment.html>


More information about the talk mailing list