Hi,
we're experiencing random build failures on the 2.0.x series
in the ShapeZipTest with the following error message:
Tests run: 10, Failures: 0, Errors: 10, Skipped: 0, Time elapsed: 1.59 sec <<< FAILURE!
testNoNativeProjection(org.geoserver.wfs.response.ShapeZipTest) Time elapsed: 0.37 sec <<< ERROR!
java.lang.NoSuchMethodError: org.geotools.feature.FeatureTypes.getFieldLength(Lorg/opengis/feature/type/AttributeDescriptor;)I
at org.geotools.data.shapefile.ShapefileDataStore.createDbaseHeader(ShapefileDataStore.java:1036)
at org.geotools.data.shapefile.ShapefileDataStore.createSchema(ShapefileDataStore.java:855)
at org.geoserver.wfs.response.ShapeZipOutputFormat.buildStore(ShapeZipOutputFormat.java:527)
at org.geoserver.wfs.response.ShapeZipOutputFormat.writeCollectionToShapefile(ShapeZipOutputFormat.java:240)
at org.geoserver.wfs.response.ShapeZipOutputFormat.write(ShapeZipOutputFormat.java:160)
at org.geoserver.wfs.response.ShapeZipTest.writeOut(ShapeZipTest.java:188)
at org.geoserver.wfs.response.ShapeZipTest.testNoNativeProjection(ShapeZipTest.java:73)
Now, recently I've expanded FeatureTypes.getFieldLength to accept a PropertyDescriptor, which is a superclass, so the change should be backwards compatible provided the client code is rebuilt.
Soo... why is this one happening? I honestly have no idea, it seems an older shapefile data store jar is being picked up that was compiled against the older API.
But I have no idea of where that might be coming from...
are we deploying snapshots somewhere other than the OpenGeo repository?
Cheers
Andrea
--
Andrea Aime
OpenGeo - http://opengeo.org
Expert service straight from the developers.
I just performed the following steps:
[dwins@anonymised.com src]$ rm -rf /home/dwins/.m2/repository/org/geotools/
[dwins@anonymised.com src]$ (svn up && cd src/ && mvn clean install) 2>&1 > gs.log
In the resulting transcript I can see the same test failures you report (I've been seeing them intermittently for a week or so, actually) and maven reports getting gt-shapefile from the opengeo repo. I guess we need to review all the possible source of the jars deployed there.
I've resolved my build issues thus far by just kicking off a geotools-2.6.x-deploy build on the hudson, so it seems that the hudson build is doing the right thing.
--
David Winslow
OpenGeo - http://opengeo.org/
On 04/14/2010 05:39 AM, Andrea Aime wrote:
Hi,
we're experiencing random build failures on the 2.0.x series
in the ShapeZipTest with the following error message:
Tests run: 10, Failures: 0, Errors: 10, Skipped: 0, Time elapsed: 1.59
sec<<< FAILURE!
testNoNativeProjection(org.geoserver.wfs.response.ShapeZipTest) Time
elapsed: 0.37 sec<<< ERROR!
java.lang.NoSuchMethodError:
org.geotools.feature.FeatureTypes.getFieldLength(Lorg/opengis/feature/type/AttributeDescriptor;)I
at
org.geotools.data.shapefile.ShapefileDataStore.createDbaseHeader(ShapefileDataStore.java:1036)
at
org.geotools.data.shapefile.ShapefileDataStore.createSchema(ShapefileDataStore.java:855)
at
org.geoserver.wfs.response.ShapeZipOutputFormat.buildStore(ShapeZipOutputFormat.java:527)
at
org.geoserver.wfs.response.ShapeZipOutputFormat.writeCollectionToShapefile(ShapeZipOutputFormat.java:240)
at
org.geoserver.wfs.response.ShapeZipOutputFormat.write(ShapeZipOutputFormat.java:160)
at org.geoserver.wfs.response.ShapeZipTest.writeOut(ShapeZipTest.java:188)
at
org.geoserver.wfs.response.ShapeZipTest.testNoNativeProjection(ShapeZipTest.java:73)
Now, recently I've expanded FeatureTypes.getFieldLength to accept a
PropertyDescriptor, which is a superclass, so the change should be
backwards compatible provided the client code is rebuilt.
Soo... why is this one happening? I honestly have no idea, it seems an
older shapefile data store jar is being picked up that was compiled
against the older API.
But I have no idea of where that might be coming from...
are we deploying snapshots somewhere other than the OpenGeo repository?
Cheers
Andrea