[mapserver-users] MapServer, OpenStreetMap, OpenLayers + Reprojection from Latlong to Mercator: a real nightmare

[thomas bonfort]

the projection blocks in your layer should be the projection of the
actual data, i.e. 4326 if you've run osm2pgsql with the -l switch. it
should *not*  be 3785 as you have put it.

--
thomas

On Wed, Mar 17, 2010 at 22:39, Mulone <mulone at rome.com> wrote:
>
> (apologies for cross-posting)
>
> Hi everybody,
>
> I'm trying to render openstreetmap data on a WMS on MapServer and I posted
> something already about this. Now it's time to summarise everything and make
> some sense of this.
>
> The system used to work with epsg:4326 (LATLONG from now on) but in our
> project we need a GoogleMaps like projection and we chose
> epsg:900913/epsg:3785 (MERC from now on).
> This is our problem.
> We need the image on the right but we're getting the one on the left, a
> flattened map:
> http://img7.imageshack.us/img7/8279/imagelw.png
>
> After a lot of work and unsucceful tests I post here the outcome, hoping
> that it will be helpful to the community - and to us :-)
>
> These are the steps I followed from the very start:
>
> 1) download osm xml file from Cloudmade (e.g. Italy.osm.bz2)
> 2) Run 900913.sql on my PostGIS (optional, it shouldn't matter).
> 3) import that file into PostGIS with command (latlong option selected):
>
> ./osm2pgsql -U userid -W -H host -d maps -p myprefix -S default.style -c -l
> italy.osm.bz2
>
> 4) At this point everything seems ok. The geometry columns have latlong as
> SRID and the tables contain geometries.
> It's possible to visualise this dataset with uDig or other tools and
> reproject it.
>
> 5) Now we want to render this stuff on MapServer. So we add the espg configs
> from
>  http://spatialreference.org/ref/epsg/3785/ to our mapserver espg file:
>
> <900913> +proj=merc +lon_0=0 +k=1 +x_0=0 +y_0=0 +a=6378137 +b=6378137
> +towgs84=0,0,0,0,0,0,0 +units=m +no_defs <>
> <3785> +proj=merc +lon_0=0 +k=1 +x_0=0 +y_0=0 +a=6378137 +b=6378137
> +towgs84=0,0,0,0,0,0,0 +units=m +no_defs <>
>
> 6) define this minimal mapfile with one layer:
> http://virtualpaste.com/82467
>
> 7) run several tests with different projections and annotate the outcome.
>
> Have a look at the report:
> http://docs.google.com/View?id=dftx6j46_166d6r38pcs
>
> Does anybody have any idea about how we could tackle this issue?
> I don't know if the problem is arising from MapServer or OpenLayers, so
> maybe you can help me design a better test to narrow the problem down.
>
> I presume a lot of people are doing something similar to us, so others might
> come across this issue.
>
> Cheers,
> Mulone
>
> -----
> Mulone
> PhD student
> --
> View this message in context: http://n2.nabble.com/MapServer-OpenStreetMap-OpenLayers-Reprojection-from-Latlong-to-Mercator-a-real-nightmare-tp4753197p4753197.html
> Sent from the Mapserver - User mailing list archive at Nabble.com.
> _______________________________________________
> mapserver-users mailing list
> mapserver-users at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/mapserver-users
>