Bill-
This fixes the ArrayIndexOutOfBoundsException and now displays
"blocky" images. I'm not sure if this is good or bad since it
effectively changes the look of all the color shaded images
in our applications. What used to be nice smooth color shading
is now blocky. Is there a way that the texture enabling can be
done on a per DataReference basis rather than the entire display?
For example, I might want some displays have nice smooth shading
and others have the blocky appearance. Could this be a new
DisplayRealType (like how Curtis added line style, etc which used
to be global) that could be added as a ConstantMap with a
DataReference?
Before and after pictures of 500 mb temperature can be found at:
http://www.unidata.ucar.edu/staff/donm/visad/colorshades.html
Don
Bill Hibbard wrote:
Using this new version of ShadowFunctionOrSetType, what once
used to (sort of) work, now doesn't. I'm using the
Radar3DCoordinateSystem and now get an ArrayOutOfBoundsException:
java.lang.ArrayIndexOutOfBoundsException
at
ucar.visad.RadarGridCoordinateSystem.toReference(RadarGridCoordinateS
ystem.java:91)
at
visad.CachingCoordinateSystem.toReference(CachingCoordinateSystem.jav
a:78)
at visad.CoordinateSystem.toReference(CoordinateSystem.java:525)
at
visad.CoordinateSystem.transformCoordinatesFreeUnits(CoordinateSystem
.java:360)
at
visad.CoordinateSystem.transformCoordinatesFreeUnits(CoordinateSystem
.java:500)
at
visad.CoordinateSystem.transformCoordinates(CoordinateSystem.java:439
)
at
visad.ShadowFunctionOrSetType.doTransform(ShadowFunctionOrSetType.jav
a:1240)
at
visad.java3d.ShadowFunctionOrSetTypeJ3D.doTransform(ShadowFunctionOrS
etTypeJ3D.java:100)
at
visad.java3d.ShadowFunctionOrSetTypeJ3D.recurseRange(ShadowFunctionOr
SetTypeJ3D.java:986)
at
visad.ShadowFunctionOrSetType.doTransform(ShadowFunctionOrSetType.jav
a:2914)
at
visad.java3d.ShadowFunctionOrSetTypeJ3D.doTransform(ShadowFunctionOrS
etTypeJ3D.java:100)
at
visad.java3d.DefaultRendererJ3D.doTransform(DefaultRendererJ3D.java:9
8)
at visad.java3d.RendererJ3D.doAction(RendererJ3D.java:177)
at visad.DisplayImpl.doAction(DisplayImpl.java:1504)
at visad.ActionImpl.run(ActionImpl.java:303)
at visad.util.ThreadPool$ThreadMinnow.run(ThreadPool.java:95)
If I revert back to the previous version, it works fine (except
for the textureMapping not working correctly). RadarGridCoordinateSystem
is attached. It looks like the 3 dimensional array it is expecting
(lat, lon, elevation) is coming in as a 2 dimensional array (lat, lon)?
I always wondered why the textureMapping didn't work for this data, so
now I guess I know why.
Oops, this problem is deeper than I thought. I have put a
possible fix at:
ftp://ftp.ssec.wisc.edu/pub/visad-2.0/untested/
and also checked it into the CVS. Let me know if it works.
Bill
