[OSM-dev] OGC standards / Sql Server 2008 / OSM... help?

Dermot McNally dermotm at gmail.com
Wed Feb 18 22:01:48 GMT 2009


Hi Brendan,

I've just looked at the piece of coastline in question and found two
errors in it (both now fixed):

* There was a duplicate node

* At a different place, the coastline overlapped itself.

I can see how these (particularly the overlap) could be confusing to
the DB. It may be worth trying it again with the new data.

Dermot

2009/2/18 brendan barrett <shogunz0r at gmail.com>:
> Hi
>
> I've just signed up to the dev mailing list, and hopefully i'll catch
> up by reading the archives. For now though I have a question and
> please don't follow up with insults if you see the names Microsoft or
> SQL Server :P
>
> I'm trying to render coastlines in an application that I am building.
> The concept is simple: find the lines, and join them end on end until
> you get an area. Now my problem is with OGC standards. I'm using Sql
> Server 2008 *ducks for cover* and more specifically the new Sql Server
> Spatial types (Geography and Geometry). Sql Server 2008 seems to be
> pretty strict with OGC standards, and I am but a beginner when it
> comes to OGC standards.
>
> The problem i'm having is with some coastlines in OSM that double back
> on themselves. Like this one:
>
> This way http://www.openstreetmap.org/browse/way/13858554 doubles back
> on itself with this node
> http://www.openstreetmap.org/browse/node/130020792 having the same
> latitude as this consecutive node
> http://www.openstreetmap.org/browse/node/130020793.
>
> I think this is the reason for the way being invalid in Sql Server's
> eyes. The MS folks have included the MakeValid() function which
> "corrects" the way to comply with OGC standards, but this changes the
> latitude and longitude of every point in the line ever so slightly,
> and that's not desirable if you are matching ways end on end. I can
> use the Node Ids to match them, but the bigger issue is that my data
> is "invalid" and I can't use other spatial methods on it.
>
> I have 4 questions:
> 1. Is this doubling back really an OGC restriction? It makes sense
> that these ways don't double back because it doesn't serve any purpose
> for a coastline (or for any line, you would expect even a path that
> doubles back to be at least some distance apart, or simply labeled two
> way).
> 2. If this is a problem, is it something that will be fixed in the
> editors / the API?
> 3. Is there anyone else here with experience with SQL Server 2008
> Spatial and OSM that might have some tips on dealing with all these
> ways?
> 4. Does PostGIS have these types of issues?
>
> I'm thinking that the only way for me to deal with this is to correct
> the data in OSM (if it's clearly incorrect) or in my database once
> imported (if there's no other solution).
>
> Thoughts?
>
> _______________________________________________
> dev mailing list
> dev at openstreetmap.org
> http://lists.openstreetmap.org/listinfo/dev
>



-- 
--------------------------------------
Iren sind menschlich




More information about the dev mailing list