Thank you for your reply.
Exactly in this moment I'm writing the answer to my problem here and on
SemanticWeb.com. In fact, during some experiments I've seen that for
each transaction (delimited by a begin() and an end()) I need to get the
default dataset and do the add operations on the new one. Instead, in my
case, I have used always the same dataset that I have initialized after
the first begin() invocation.
The inconsistency arises from this kind of problem. So in order to
solve, I've assigned the result of getDefaultModel() to my datasetModel
reference after the begin() invocation in order to solve it.
I hope that this post will help others.
Cheers,
Alessandro Suglia
On 06/27/14 18:38, Andy Seaborne wrote:
Hi there,
What does PropertiesManager do? That seems critical here.
You should be using dataset transactions, ReadWrite.WRITE for an update.
Which version of jena is this?
> Inconsistency:
> base.allocOffset() = 1982292 : allocOffset = 0
That usually means that at some point in the past, either a
non-transactional update creash, or the app didn't sync the database
before exiting. Another possible cause is two JVMs accress the same
database at the same time.
Andy
On 27/06/14 12:22, Alessandro Suglia wrote:
Good morning to all, I'm having some trouble using apache Jena TDB API,
in particular when I have to add a lot of statement to my model.
I need to save the result which I get from the dbpedia endpoint for
about two thousand entities. For each entity I get about ten properties
that I need for my task. So, when I execute the query, I loop through
all the result set and do multiple add operations for each statement in
the result set.
This is the code that I use in order to save the properties for a single
subject:
|public void saveResourceProperties(String resourceURI,
Collection<String> specificProp, PropertiesManager propManager) throws
Exception {
String proprVariable = "?prop",
valueVariable = "?value",
fixedURI = "<" + resourceURI + ">",
queryProp = "select " + proprVariable + " " + valueVariable
+ " where {\n" +
fixedURI + " " + proprVariable + " " +
valueVariable +
" values " + proprVariable + "{" +
formatPropertiesList(specificProp) + "} \n" +
"}";
Query query = QueryFactory.create(queryProp);
QueryExecution qexec = null;
try {
qexec = QueryExecutionFactory.sparqlService(endpoint, query);
ResultSet resultSet = qexec.execSelect();
QuerySolution currSolution;
// invokes dataset.begin(ReadWrite.Write);
propManager.start(true);
while (resultSet.hasNext()) {
currSolution = resultSet.nextSolution();
writer.write(resourceURI + "\t" +
currSolution.getResource(proprVariable).toString() +
"\t" + currSolution.get(valueVariable).toString() +
"\n");
// prepare the statement using the resource URI and the
current solution
// and adds it using add() method
propManager.addSolution(currSolution, resourceURI);
}
// I commit the results calling dataset.commit()
propManager.commitChanges();
}
finally {
if (qexec != null)
qexec.close();
// calls dataset.end() in order to end the current transaction
propManager.closeManager();
}
}
|
Everything goes ok, but at the end of the process I receive this
exception (it happens each time that I call this method): Inconsistency:
base.allocOffset() = 1982292 : allocOffset = 0 After that, no file are
stored and for this reason I'm not able to read nothing.
I've tried to execute all the commit in a single, huge transaction, but
I receive the same error and the program is not able to store TDB index
files.
How can I manage this kind of situation? Can you suggest to me an
efficient way to do that? I'm getting mad...
Thank you in advance.
Alessandro Suglia
