[Geoserver-users] Some Missing Polygons

David_Collins · June 22, 2009, 2:43am

Andrea,

Thanks for the clarification - I understand now.

I have been investigating the transparent layer further.

FYI, this is the URL - at low zooms Geoserver sends a raster image, at
high zooms vector information is sent to GE ..
http://dpi12345:8080/geoserver/wms/kml?layers=Geology&mode=refresh&kmscore=50&kmattr=true&

I changed 'mode' to 'download', but this makes no difference.

In the Geological map there are 19,000 polygons having 51 colours. I
have only detected the missing colour for one type of Geology and that
has a 'polycolour' field of 12. This is not the lowest value - there
is a polycolour value of 5. All other polycolour values are higher.
When Geoserver sends a raster image, ALL colours appear. There is
only a problem when Geoserver sends vector output to GE.

Another note - some polygons with polycolour 12 appear properly,
others do not. It seems to be the large long polygons that have the
problem. The Geology that has the problem is river sediment so, as
you can imagine, the shapes are long as they follow rivers.

I am wondering if there is a problem with polygons that extend a long
way out of the area that is zoomed in on?

I am getting these errors to the console - are these relevant?
22 Jun 22:07:10 WARN [geotools.rendering] - Assuming rendering buffer
= 0, but estimation is not accurate, you may want to set a buffer
manually
class org.vfny.geoserver.global.GeoServerFeatureLocking

I have also attached an image.

Regards,
David

On Fri, Jun 19, 2009 at 5:42 PM, Andrea Aime<aaime@anonymised.com> wrote:

David Collins ha scritto:
Andrea,

Re: 'if there are more than 20 rules ... no filter (besides the bbox)
will be sent to the database'

I think I might not be understanding what you have said. With
Geological maps, we need to use many more than 20 colours - the
Geological map that I am using for testing Geoserver at the moment has
51 colours - ie. my SLD has 51 rules based on a field called
'polycolour' and when I view it as a rasterised image in Google Earth
or uDig all the colours seem to show OK. I do have a problem, though,
when I zoom in and get vector output in GE - one of the colours is
missing (ie. transparent) but it is not the 21st rule - it is my 2nd
rule ...
  &lt;Rule&gt;
     &lt;Name&gt;12&lt;/Name&gt;
     &lt;Title&gt;12&lt;/Title&gt;
     &lt;ogc:Filter&gt;
       &lt;ogc:PropertyIsEqualTo&gt;
         &lt;ogc:PropertyName&gt;POLYCOLOUR&lt;/ogc:PropertyName&gt;
         &lt;ogc:Literal&gt;12&lt;/ogc:Literal&gt;
       &lt;/ogc:PropertyIsEqualTo&gt;
     &lt;/ogc:Filter&gt;
     &lt;PolygonSymbolizer&gt;
       &lt;Fill&gt;
         &lt;CssParameter name=&quot;fill&quot;&gt;\#FFFFCC&lt;/CssParameter&gt;
         &lt;CssParameter name=&quot;fill\-opacity&quot;&gt;1&lt;/CssParameter&gt;
       &lt;/Fill&gt;
     &lt;/PolygonSymbolizer&gt;
   &lt;/Rule&gt;
Can you please clarify what you mean about the 20 rule maximum?
When hitting the data storage we can decide to OR all the rule
filters into a giant query, like:

select ... from table where geom in bbox and (prop = val1 or prop = val2
or prop = val3 or ...)

or just assume that you're not really trying to filter your data but
just coloring each with a different color, and thus a query like:

select ... from table where geom in bbox

is just as good. The threshold between the two behaviors is at 20
rules. It does not mean you cannot have more than 20 rules,
just means after 20 the query issued against the server will change.

As to why the second color is not showing up in GE, might it be
because the superoverlay code is deciding that those feature are
less important and thus you have to zoom in more?

Try adding &mode=download to your kml link, does it work any better?
Cheers
Andrea

--
Andrea Aime
OpenGeo - http://opengeo.org
Expert service straight from the developers.

David_Collins · June 24, 2009, 3:27am

Hello,

I have looked at this problem further, and it seems that it is just
one or two large polygons that are 'missing' in GE. This polygon(s)
takes up possibly as much as 5% of the whole area, so is much larger
than the other 19,000 polygons.
(It has less than 36,000 points, though.)

I am now trying to see if the problem is with Google Earth or Geoserver.

As mentioned below, the polygon is only missing when the vector data
is sent to GE (not raster).

The resultant URL in the 'Places' area in the left pane of GE is ..
http://dpi12345:8080/geoserver/wms?height=1024&width=1024&layers=topp%3AGeology&request=GetMap&service=wms&styles=geology&format_options=SUPEROVERLAY%3Afalse%3BKMPLACEMARK%3Afalse%3BKMSCORE%3A50%3BKMATTR%3Atrue%3B&srs=EPSG%3A4326&format=application%2Fvnd.google-earth.kmz%2Bxml&transparent=false&version=1.1.1

If I use uDig, I do not get the problem with either WMS of WFS, but I
understand that WMS is always sending raster to uDig - it that correct
?

If so, is there a program apart from GE where I can get WMS to send
vector output to a client ?

Regards,
David Collins

---------- Forwarded message ----------
From: David Collins <david.8.collins@anonymised.com>
Date: Mon, Jun 22, 2009 at 12:43 PM
Subject: Re: [Geoserver-users] Some Missing Polygons
To: geoserver-users@lists.sourceforge.net