HI Andrea: The WMS Lab code was donated; I put some time into it last year before pulling it from the tutorials (for the reasons you mention).
There is an earlier tutorial from 2006 I was wanting to replace - which simply showed how to use the WMS API; retrieve a list of layers and make a GetMap request (no attempt at panning or anything). On the email list last year someone mentioned a JMapPane alternative that used a background thread to draw; leaving the user interface thread free to pan around quickly. I was hoping they would donate that code ... but apparently they never got permission from their company to. That would be the approach I would take; a user interface that takes responsibility for running a background thread - and was able to drive WMS and the GeoTools renderer. Wrapping the resulting image up as a GridCoverage is actually useful for getting everything to line up at the end of the day. It is also a clean way to handle tile set results. I can port the WMS-C code into geotools for you - indeed I would like to combine it with the existing WMS module. Emily did a nice job of parsing the vendor specific GetCapabilities document; and the fetching of tiles is already nicely isolated (with a memory and a disk strategy for caching). It uses the Eclipse Jobs API so it would also need to be migrate to using a Future. Jody On 07/03/2010, at 6:49 AM, Andrea Aime wrote: > Hi, > today I had a look at WMSLab to see what was the issue reported > by a user on the users list. > > After playing a bit with it, I'm wondering if it actually ever worked. > I have a patch that solves some of the issues, it's attached for > review. > > Issues I've found: > - the raster symbolizer is created that looks into an attribute named > "raster" for the grid. However the wrapped coverages have been called > "grid" for a long time (I went back in the history quite a bit > and did not find a time in which they were called "raster" instead). > Long story short, with this difference it could not have worked, > not now, neither two years ago > - the WMSMapLayer seems an unfinished work in progress. > It assumes it can turn loose a runnable that will fetch them map, > and that it will return in time for the getFeatureSource() call. > Purely assumes, I mean, there is no synchronization at all. > If it eve worked, it did by accident. > Also the class was filled with System.out > I modified the class to use Future in order to wait for the > return value and turned the System.out into LOGGER calls > - the WMSMapLayer has to play tricks in order to issue the > GetMap request, in particular it requires the mapBoundsChanged > to be called, but that happens only if whoever adds the > layer also registers it as a bounds listener to the map. > I modified the WMSLab to do so. > > With all the modifications above the example actually turns > out images and you can pan and zoom. > > There are however a number of issues still unsolved, which > I hope someone else might tackle: > - the map is always distorted. This happens because the > WMSMapLayer always assumes a painting area of 200x400, > and someone has to change those values in the various > setter of the WMSMapLayer dynamically... ugh! > - the feature info request does not work because the > GUI assumes the layer is a raster. Well, it's a fake > one, the feature info tool needs to be modified > to be aware of that > - the WMSMapLayer uses a thread pool which is not bounded, > meaning it will do any amount of requests in parallel. > This is very bad behaviour, all HTTP clients are supposed > to avoid making more than 4-6 requests in parallel against > the same server to avoid overwhelm it. Years ago browsers > never did more than 2 in parallel, nowadays Firefox does > 6, but never more (unless you install some accelerator) > - the reprojection simply does not work, but I did not > investigate on the why > > Thinking about it, I believe it would be much better if > WMSMapLayer was implemented in a different way. > My best bet would be on a layer that wraps a custom > coverage reader, the reader being in fact something > that delegates the request to the WMS server _when_ > it's time to render that layer. > By doing so the reader would get all the necessary > hints, the actual geographic area to be painted and > also the on screen rectangle to be generated. > It would also not need any kind of synchronization, because > it would do the request on demand. > The downside is that it would not be able to make > requests in parallel as it does now (since now the request > starts when the area of interest of the map changes, > not when the getFeatureSource() method is called). > > If people are interested in the above approach I > can try to implement it. I'll do eventually anyways > when I start implementing WMS cascading for GeoServer, > but that will happen only in a couple of months > time > > Cheers > Andrea > > > -- > Andrea Aime > OpenGeo - http://opengeo.org > Expert service straight from the developers. > Index: > modules/library/main/src/main/java/org/geotools/styling/RasterSymbolizerImpl.java > =================================================================== > --- > modules/library/main/src/main/java/org/geotools/styling/RasterSymbolizerImpl.java > (revisione 34976) > +++ > modules/library/main/src/main/java/org/geotools/styling/RasterSymbolizerImpl.java > (copia locale) > @@ -60,7 +60,7 @@ > } > > public RasterSymbolizerImpl(FilterFactory factory, Description desc, > String name, Unit<Length> uom, OverlapBehavior behavior) { > - super(name, desc, "raster", uom); > + super(name, desc, "grid", uom); > this.filterFactory = factory; > this.opacity = filterFactory.literal(1.0); > this.overlap = filterFactory.literal(OverlapBehavior.RANDOM); > Index: modules/extension/wms/src/main/java/org/geotools/map/WMSMapLayer.java > =================================================================== > --- modules/extension/wms/src/main/java/org/geotools/map/WMSMapLayer.java > (revisione 34976) > +++ modules/extension/wms/src/main/java/org/geotools/map/WMSMapLayer.java > (copia locale) > @@ -14,7 +14,6 @@ > import java.awt.event.ComponentListener; > import java.awt.image.BufferedImage; > import java.io.BufferedReader; > -import java.io.IOException; > import java.io.InputStream; > import java.io.InputStreamReader; > import java.net.URL; > @@ -23,6 +22,7 @@ > import java.util.Set; > import java.util.concurrent.ExecutorService; > import java.util.concurrent.Executors; > +import java.util.concurrent.Future; > import java.util.logging.Level; > import java.util.logging.Logger; > > @@ -38,13 +38,13 @@ > import org.geotools.data.wms.WebMapServer; > import org.geotools.data.wms.request.GetMapRequest; > import org.geotools.factory.CommonFactoryFinder; > -import org.geotools.factory.FactoryRegistryException; > import org.geotools.geometry.GeneralEnvelope; > import org.geotools.geometry.jts.ReferencedEnvelope; > import org.geotools.map.event.MapBoundsEvent; > import org.geotools.map.event.MapBoundsListener; > import org.geotools.map.event.MapLayerEvent; > import org.geotools.referencing.CRS; > +import org.geotools.renderer.GTRenderer; > import org.geotools.resources.coverage.FeatureUtilities; > import org.geotools.styling.FeatureTypeStyle; > import org.geotools.styling.RasterSymbolizer; > @@ -54,9 +54,14 @@ > import org.geotools.styling.Symbolizer; > import org.opengis.feature.simple.SimpleFeature; > import org.opengis.feature.simple.SimpleFeatureType; > -import org.opengis.referencing.FactoryException; > import org.opengis.referencing.crs.CoordinateReferenceSystem; > > +/** > + * Integrates a WMS layer into a {...@link MapContext}. For this layer to > work it is > + * crucial that the layer is added as a {...@link MapBoundsListener} to the > MapContext and > + * that the MapContext area of interest is kept in synch with whatever is > provided to > + * the {...@link GTRenderer} paint calls > + */ > public class WMSMapLayer extends DefaultMapLayer implements MapLayer, > MapBoundsListener, > ComponentListener { > /** The logger for the map module. */ > @@ -88,6 +93,8 @@ > private String exceptions = "application/vnd.ogc.se_inimage"; > > private static ExecutorService executor = Executors.newCachedThreadPool(); > + > + private static Future<?> future; > > Runnable getmap = new Runnable() { > public void run() { > @@ -104,8 +111,27 @@ > this.layer = layer; > this.wms = wms; > setDefaultStyle(); > - executor.execute(getmap); > } > + > + @Override > + public FeatureSource<SimpleFeatureType, SimpleFeature> > getFeatureSource() { > + if(featureSource == null) { > + if(future == null) { > + future = executor.submit(getmap); > + } > + } > + > + if(future != null) { > + try { > + future.get(); > + future = null; > + } catch(Exception e) { > + throw new RuntimeException("Failed to perform the GetMap > request"); > + } > + } > + > + return featureSource; > + } > > public void setDefaultStyle() { > RasterSymbolizer symbolizer = factory.createRasterSymbolizer(); > @@ -154,7 +180,6 @@ > GetMapRequest mapRequest = wms.createGetMapRequest(); > mapRequest.addLayer(layer); > > - // System.out.println(width + " " + height); > mapRequest.setDimensions(getWidth(), getHeight()); > mapRequest.setFormat("image/png"); > > @@ -184,14 +209,13 @@ > requestBBox = getBounds(); > } > > - // System.out.println(bbox.toString()); > mapRequest.setSRS(srsName); > mapRequest.setBBox(requestBBox); > mapRequest.setTransparent(transparent); > > URL request = mapRequest.getFinalURL(); > > - System.out.println(request.toString()); > + LOGGER.fine(request.toString()); > InputStream is = null; > > try { > @@ -201,26 +225,24 @@ > > if (type.equalsIgnoreCase("image/png")) { > BufferedImage image = ImageIO.read(is); > - System.out.println("get map completed"); > + LOGGER.fine("GetMap completed"); > grid = gcf.create(layer.getTitle(), image, requestBBox); > - // System.out.println("fetched new grid"); > - // if (featureSource == null) > featureSource = > DataUtilities.source(FeatureUtilities.wrapGridCoverage(grid)); > if( viewport == null ){ > - System.out.println("get map complete but viewport not > yet established"); > + LOGGER.fine("Get map complete but viewport not yet > established"); > } > else { > fireMapLayerListenerLayerChanged(new MapLayerEvent(this, > MapLayerEvent.DATA_CHANGED)); > } > } else { > - System.out.println("error content type is " + type); > + LOGGER.severe("GetMap returned an error, content type is " + > type); > > if (StringUtils.contains(type, "text") || > StringUtils.contains(type, "xml")) { > String line = ""; > BufferedReader br = new BufferedReader(new > InputStreamReader(is)); > > while ((line = br.readLine()) != null) { > - System.out.println(line); > + LOGGER.severe(line); > } > } > } > @@ -234,12 +256,10 @@ > public void mapBoundsChanged(MapBoundsEvent event) { > viewport = ((MapContext) event.getSource()).getAreaOfInterest(); > > - // System.out.println("old:" + bbox + "\n" + "new:" > - // + event.getOldAreaOfInterest()); > if (!viewport.equals(event.getOldAreaOfInterest())) { > - System.out.println("bbox changed - fetching new grid"); > + LOGGER.fine("bbox changed - fetching new grid"); > > - executor.execute(getmap); > + future = executor.submit(getmap); > } > } > > Index: demo/example/src/main/java/org/geotools/demo/WMSLab.java > =================================================================== > --- demo/example/src/main/java/org/geotools/demo/WMSLab.java (revisione > 34976) > +++ demo/example/src/main/java/org/geotools/demo/WMSLab.java (copia locale) > @@ -44,6 +44,7 @@ > for( Layer wmsLayer : wmsLayers ){ > WMSMapLayer displayLayer = new WMSMapLayer(wms, wmsLayer ); > mapcontext.addLayer( displayLayer ); > + mapcontext.addMapBoundsListener(displayLayer); > } > // Now display the map > JMapFrame.showMap(mapcontext); > ------------------------------------------------------------------------------ > Download Intel® Parallel Studio Eval > Try the new software tools for yourself. Speed compiling, find bugs > proactively, and fine-tune applications for parallel performance. > See why Intel Parallel Studio got high marks during beta. > http://p.sf.net/sfu/intel-sw-dev_______________________________________________ > Geotools-devel mailing list > Geotools-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/geotools-devel ------------------------------------------------------------------------------ Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev _______________________________________________ Geotools-devel mailing list Geotools-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/geotools-devel
