On Thu, 2009-09-10 at 23:13 +0100, Jon Burgess wrote:
> On Thu, 2009-09-10 at 17:48 -0400, James McManus wrote:
> > Hi - I'm having an unusual problem with some maps I produced with
> > Mapnik. With some choropleth maps I'm getting incorrect output. This
> > can best be seen by comparing maps I've produced using Mapnik:
> > http://www.piedmontgeographic.com/wakesearch.html and Mapserver:
> > http://www.piedmontgeographic.com/example1.html
> >
> > If you look at the center of the map, you will see a tract (tractid
> > 051600) that has a different color value between the Mapnik and
> > Mapserver produced images. If you activate the searcher and select
> > this tract you will see that the avg_r_tax_val_0906 is 670473.65, wich
> > means the Mapserver image is correct. How is the happening? I've
> > examined the mapfile many times, and it is correct. Most of the
> > tracts are correct. Only a few come out incorrect!
>
> It would be worth checking that the polygons describing the missing
> tracts are valid geometries, e.g.
>
> => select isvalid(the_geom) from table_name where tract_id=...
>
> You should also verify that they are defined in a clockwise direction,
> the ST_ForceRHR()[1] function should be able to do this for you.
>
> If those do not help then you may need to narrow down the data to see if
> there is anything else which looks odd about those missing polygons.
Something else you might want to watch out for is broken filters. You
could add an <ElseFilter> to your style with a distinctive colour. If
the missing areas show with this then the filter matching for the data
is broken. If they do not, then you may want to capture the queries in
postgres to see what mapnik is asking for.
A few issues have been seen where filters fail dues to formatting
issues, like fields with extra padding spaces. Something similar could
happen with integer vs non-integer values.
Jon
_______________________________________________
Mapnik-users mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/mapnik-users
