com.bigdata.relation.rule.eval.AbstractJoinNexus Maven / Gradle / Ivy
/**
Copyright (C) SYSTAP, LLC DBA Blazegraph 2006-2016. All rights reserved.
Contact:
SYSTAP, LLC DBA Blazegraph
2501 Calvert ST NW #106
Washington, DC 20008
[email protected]
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; version 2 of the License.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
/*
* Created on Aug 20, 2010
*/
package com.bigdata.relation.rule.eval;
import java.util.Iterator;
import java.util.Map;
import java.util.concurrent.TimeUnit;
import org.apache.log4j.Logger;
import com.bigdata.bop.Constant;
import com.bigdata.bop.IBindingSet;
import com.bigdata.bop.IConstant;
import com.bigdata.bop.IConstraint;
import com.bigdata.bop.IElement;
import com.bigdata.bop.IPredicate;
import com.bigdata.bop.IVariable;
import com.bigdata.bop.IVariableOrConstant;
import com.bigdata.bop.bindingSet.ListBindingSet;
import com.bigdata.bop.joinGraph.DefaultRangeCountFactory;
import com.bigdata.bop.joinGraph.IEvaluationPlanFactory;
import com.bigdata.bop.joinGraph.IRangeCountFactory;
import com.bigdata.btree.keys.DelegateSortKeyBuilder;
import com.bigdata.btree.keys.ISortKeyBuilder;
import com.bigdata.config.Configuration;
import com.bigdata.config.IValidator;
import com.bigdata.config.IntegerValidator;
import com.bigdata.config.LongValidator;
import com.bigdata.io.IStreamSerializer;
import com.bigdata.io.SerializerUtil;
import com.bigdata.journal.IIndexManager;
import com.bigdata.journal.Journal;
import com.bigdata.journal.TemporaryStore;
import com.bigdata.mdi.PartitionLocator;
import com.bigdata.relation.AbstractResource;
import com.bigdata.relation.IMutableRelation;
import com.bigdata.relation.IRelation;
import com.bigdata.relation.RelationFusedView;
import com.bigdata.relation.accesspath.AccessPath;
import com.bigdata.relation.accesspath.BlockingBuffer;
import com.bigdata.relation.accesspath.IAccessPath;
import com.bigdata.relation.accesspath.IBlockingBuffer;
import com.bigdata.relation.accesspath.IBuffer;
import com.bigdata.relation.accesspath.IElementFilter;
import com.bigdata.relation.accesspath.UnsynchronizedArrayBuffer;
import com.bigdata.relation.locator.IResourceLocator;
import com.bigdata.relation.rule.IProgram;
import com.bigdata.relation.rule.IRule;
import com.bigdata.relation.rule.IStep;
import com.bigdata.service.AbstractScaleOutFederation;
import com.bigdata.service.DataService;
import com.bigdata.service.IBigdataFederation;
import com.bigdata.service.ndx.IClientIndex;
import com.bigdata.striterator.ChunkedConvertingIterator;
import com.bigdata.striterator.DistinctFilter;
import com.bigdata.striterator.IChunkedOrderedIterator;
/**
* Base implementation for {@link IJoinNexus}
*
* @author Bryan Thompson
* @version $Id$
*/
abstract public class AbstractJoinNexus implements IJoinNexus {
private final static transient Logger log = Logger.getLogger(AbstractJoinNexus.class);
private final IJoinNexusFactory joinNexusFactory;
protected final IIndexManager indexManager;
/** Note: cached. */
protected final IResourceLocator resourceLocator;
private final ActionEnum action;
private final long writeTimestamp;
protected final long readTimestamp;
protected final int chunkCapacity;
protected final int chunkOfChunksCapacity;
private final boolean forceSerialExecution;
private final int maxParallelSubqueries;
private final int fullyBufferedReadThreshold;
private final long chunkTimeout;
/**
* The {@link TimeUnit}s in which the {@link #chunkTimeout} is measured.
*/
private static final TimeUnit chunkTimeoutUnit = TimeUnit.MILLISECONDS;
final public int getChunkOfChunksCapacity() {
return chunkOfChunksCapacity;
}
final public int getChunkCapacity() {
return chunkCapacity;
}
final public int getFullyBufferedReadThreshold() {
return fullyBufferedReadThreshold;
}
final public String getProperty(final String name, final String defaultValue) {
// @todo pass namespace in with the IJoinNexusFactory?
return Configuration.getProperty(indexManager,
joinNexusFactory.getProperties(), null/* namespace */, name,
defaultValue);
}
final public T getProperty(final String name, final String defaultValue,
final IValidator validator) {
// @todo pass namespace in with the IJoinNexusFactory?
return Configuration.getProperty(indexManager,
joinNexusFactory.getProperties(), null/* namespace */, name,
defaultValue, validator);
}
protected final int solutionFlags;
protected final IElementFilter filter;
public IElementFilter getSolutionFilter() {
return filter == null ? null : new SolutionFilter(filter);
}
/**
* The factory for rule evaluation plans.
*/
protected final IEvaluationPlanFactory planFactory;
/**
* @todo caching for the same relation and database state shared across join
* nexus instances.
*/
private final IRangeCountFactory rangeCountFactory = new DefaultRangeCountFactory(
this);
private final IRuleStatisticsFactory ruleStatisticsFactory = new IRuleStatisticsFactory() {
public RuleStats newInstance(IStep step) {
return new RuleStats(step);
}
public RuleStats newInstance(IRuleState ruleState) {
return new RuleStats(ruleState);
}
};
/**
* @param joinNexusFactory
* The object used to create this instance and which can be used
* to create other instances as necessary for distributed rule
* execution.
* @param indexManager
* The object used to resolve indices, relations, etc.
*/
protected AbstractJoinNexus(final IJoinNexusFactory joinNexusFactory,
final IIndexManager indexManager) {
if (joinNexusFactory == null)
throw new IllegalArgumentException();
if (indexManager == null)
throw new IllegalArgumentException();
this.joinNexusFactory = joinNexusFactory;
this.indexManager = indexManager;
this.resourceLocator = indexManager.getResourceLocator();
this.action = joinNexusFactory.getAction();
this.writeTimestamp = joinNexusFactory.getWriteTimestamp();
this.readTimestamp = joinNexusFactory.getReadTimestamp();
forceSerialExecution = Boolean.parseBoolean(getProperty(
AbstractResource.Options.FORCE_SERIAL_EXECUTION,
AbstractResource.Options.DEFAULT_FORCE_SERIAL_EXECUTION));
maxParallelSubqueries = getProperty(
AbstractResource.Options.MAX_PARALLEL_SUBQUERIES,
AbstractResource.Options.DEFAULT_MAX_PARALLEL_SUBQUERIES,
IntegerValidator.GTE_ZERO);
chunkOfChunksCapacity = getProperty(
AbstractResource.Options.CHUNK_OF_CHUNKS_CAPACITY,
AbstractResource.Options.DEFAULT_CHUNK_OF_CHUNKS_CAPACITY,
IntegerValidator.GT_ZERO);
chunkCapacity = getProperty(AbstractResource.Options.CHUNK_CAPACITY,
AbstractResource.Options.DEFAULT_CHUNK_CAPACITY,
IntegerValidator.GT_ZERO);
chunkTimeout = getProperty(AbstractResource.Options.CHUNK_TIMEOUT,
AbstractResource.Options.DEFAULT_CHUNK_TIMEOUT,
LongValidator.GTE_ZERO);
fullyBufferedReadThreshold = getProperty(
AbstractResource.Options.FULLY_BUFFERED_READ_THRESHOLD,
AbstractResource.Options.DEFAULT_FULLY_BUFFERED_READ_THRESHOLD,
IntegerValidator.GT_ZERO);
this.solutionFlags = joinNexusFactory.getSolutionFlags();
this.filter = joinNexusFactory.getSolutionFilter();
this.planFactory = joinNexusFactory.getEvaluationPlanFactory();
}
public IRuleStatisticsFactory getRuleStatisticsFactory() {
return ruleStatisticsFactory;
}
public IJoinNexusFactory getJoinNexusFactory() {
return joinNexusFactory;
}
public IRangeCountFactory getRangeCountFactory() {
return rangeCountFactory;
}
final public boolean forceSerialExecution() {
if (log.isInfoEnabled())
log.info("forceSerialExecution="+forceSerialExecution);
return forceSerialExecution;
}
final public int getMaxParallelSubqueries() {
return maxParallelSubqueries;
}
final public ActionEnum getAction() {
return action;
}
final public long getWriteTimestamp() {
return writeTimestamp;
}
final public long getReadTimestamp() {
return readTimestamp;
}
/**
* The head relation is what we write on for mutation operations and is also
* responsible for minting new elements from computed {@link ISolution}s.
* This method depends solely on the name of the head relation and the
* timestamp of interest for the view.
*/
public IRelation getHeadRelationView(final IPredicate pred) {
if (pred.getRelationCount() != 1)
throw new IllegalArgumentException();
final String relationName = pred.getOnlyRelationName();
final long timestamp = (getAction().isMutation() ? getWriteTimestamp()
: getReadTimestamp(/*relationName*/));
return (IRelation) resourceLocator.locate(relationName, timestamp);
}
@SuppressWarnings("unchecked")
public IRelation getTailRelationView(final IPredicate pred) {
final int nsources = pred.getRelationCount();
if (nsources == 1) {
return (IRelation) resourceLocator.locate(pred
.getOnlyRelationName(), getReadTimestamp());
} else if (nsources == 2) {
final IRelation relation0 = (IRelation) resourceLocator.locate(
pred.getRelationName(0), readTimestamp);
final IRelation relation1 = (IRelation) resourceLocator.locate(
pred.getRelationName(1), readTimestamp);
return new RelationFusedView(relation0, relation1).init();
} else {
throw new UnsupportedOperationException();
}
}
// /**
// * @deprecated by {@link #getTailAccessPath(IRelation, IPredicate)}
// *
// * @see #getTailAccessPath(IRelation, IPredicate).
// */
// public IAccessPath getTailAccessPath(final IPredicate predicate) {
//
// // Resolve the relation name to the IRelation object.
// final IRelation relation = getTailRelationView(predicate);
//
// return getTailAccessPath(relation, predicate);
//
// }
public IAccessPath getTailAccessPath(final IRelation relation,
final IPredicate predicate) {
return relation.getAccessPath(indexManager/* localIndexManager */,
relation.getKeyOrder(predicate), predicate);
// if (predicate.getPartitionId() != -1) {
//
// /*
// * Note: This handles a read against a local index partition. For
// * scale-out, the [indexManager] will be the data service's local
// * index manager.
// *
// * Note: Expanders ARE NOT applied in this code path. Expanders
// * require a total view of the relation, which is not available
// * during scale-out pipeline joins. Likewise, the [backchain]
// * property will be ignored since it is handled by an expander.
// */
//
// return ((AbstractRelation) relation)
// .getAccessPathForIndexPartition(indexManager, predicate);
//
// }
//
// // Find the best access path for the predicate for that relation.
// final IAccessPath accessPath = relation.getAccessPath(predicate);
//
// // Note: No expander's for bops, at least not right now.
//// final ISolutionExpander expander = predicate.getSolutionExpander();
////
//// if (expander != null) {
////
//// // allow the predicate to wrap the access path
//// accessPath = expander.getAccessPath(accessPath);
////
//// }
//
// // return that access path.
// return accessPath;
}
public Iterator locatorScan(
final AbstractScaleOutFederation fed,
final IPredicate predicate) {
final long timestamp = getReadTimestamp();
// Note: assumes that we are NOT using a view of two relations.
final IRelation relation = (IRelation) fed.getResourceLocator().locate(
predicate.getOnlyRelationName(), timestamp);
/*
* Find the best access path for the predicate for that relation.
*
* Note: All we really want is the [fromKey] and [toKey] for that
* predicate and index. This MUST NOT layer on expanders since the
* layering also hides the [fromKey] and [toKey].
*/
@SuppressWarnings("unchecked")
final AccessPath accessPath = (AccessPath) relation
.getAccessPath((IPredicate)predicate);
// Note: assumes scale-out (EDS or JDS).
final IClientIndex ndx = (IClientIndex) accessPath.getIndex();
/*
* Note: could also be formed from relationName + "." +
* keyOrder.getIndexName(), which is cheaper unless the index metadata
* is cached.
*/
final String name = ndx.getIndexMetadata().getName();
return fed.locatorScan(name, timestamp, accessPath.getFromKey(),
accessPath.getToKey(), false/* reverse */);
}
final public IIndexManager getIndexManager() {
return indexManager;
}
@SuppressWarnings("unchecked")
final public boolean bind(final IRule rule, final int index,
final Object e, final IBindingSet bindings) {
// propagate bindings from the visited object into the binding set.
copyValues((IElement) e, rule.getTail(index), bindings);
// verify constraints.
return rule.isConsistent(bindings);
}
final public boolean bind(final IPredicate pred,
final IConstraint[] constraints, final Object e,
final IBindingSet bindings) {
// propagate bindings from the visited object into the binding set.
copyValues((IElement) e, pred, bindings);
if (constraints != null) {
// verify constraint.
return isConsistent(constraints, bindings);
}
// no constraint.
return true;
}
/**
* Check constraints.
*
* @param constraints
* @param bindingSet
*
* @return true iff the constraints are satisfied.
*/
private boolean isConsistent(final IConstraint[] constraints,
final IBindingSet bindingSet) {
for (int i = 0; i < constraints.length; i++) {
final IConstraint constraint = constraints[i];
if (!constraint.accept(bindingSet)) {
if (log.isDebugEnabled()) {
log.debug("Rejected by "
+ constraint.getClass().getSimpleName() + " : "
+ bindingSet);
}
return false;
}
}
return true;
}
@SuppressWarnings("unchecked")
final private void copyValues(final IElement e, final IPredicate pred,
final IBindingSet bindingSet) {
for (int i = 0; i < pred.arity(); i++) {
final IVariableOrConstant t = pred.get(i);
if (t.isVar()) {
final IVariable var = (IVariable) t;
final Constant newval = new Constant(e.get(i));
bindingSet.set(var, newval);
}
}
}
final public ISolution newSolution(final IRule rule,
final IBindingSet bindingSet) {
final Solution solution = new Solution(this, rule, bindingSet);
if (log.isDebugEnabled()) {
log.debug(solution.toString());
}
return solution;
}
final public int solutionFlags() {
return solutionFlags;
}
/**
* FIXME Custom serialization for solution sets, especially since there
* tends to be a lot of redundancy in the data arising from how bindings are
* propagated during JOINs.
*
* @todo We can sort the {@link ISolution}s much like we already do for
* DISTINCT or intend to do for SORT and use the equivalent of leading
* key compression to reduce IO costs (or when they are SORTed we
* could leverage that to produce a more compact serialization).
*
* @see SPOSolutionSerializer (needs to be written).
*/
public IStreamSerializer getSolutionSerializer() {
return SerializerUtil.STREAMS;
}
/**
* FIXME Custom serialization for binding sets, especially since there tends
* to be a lot of redundancy in the data arising from how bindings are
* propagated during JOINs.
*
* @todo We can sort the {@link ISolution}s much like we already do for
* DISTINCT or intend to do for SORT and use the equivalent of leading
* key compression to reduce IO costs (or when they are SORTed we
* could leverage that to produce a more compact serialization).
*
* @see SPOBindingSetSerializer, which has not been finished.
*/
public IStreamSerializer getBindingSetSerializer() {
return SerializerUtil.STREAMS;
}
final public IBindingSet newBindingSet(final IRule rule) {
final IBindingSet constants = rule.getConstants();
final int nconstants = constants.size();
final IBindingSet bindingSet =
new ListBindingSet()
// new ArrayBindingSet(rule.getVariableCount()+ nconstants)
;
if (nconstants > 0) {
/*
* Bind constants declared by the rule before returning the binding
* set to the caller.
*/
final Iterator> itr = constants
.iterator();
while(itr.hasNext()) {
final Map.Entry entry = itr.next();
bindingSet.set(entry.getKey(), entry.getValue());
}
}
return bindingSet;
}
final public IRuleTaskFactory getRuleTaskFactory(final boolean parallel,
final IRule rule) {
if (rule == null)
throw new IllegalArgumentException();
// is there a task factory override?
IRuleTaskFactory taskFactory = rule.getTaskFactory();
if (taskFactory == null) {
// no, use the default factory.
taskFactory = joinNexusFactory.getDefaultRuleTaskFactory();
}
/*
* Note: Now handled by the MutationTask itself.
*/
// if (getAction().isMutation() && (!parallel || forceSerialExecution())) {
//
// /*
// * Tasks for sequential mutation steps are always wrapped to ensure
// * that the thread-safe buffer is flushed onto the mutable relation
// * after each rule executes. This is necessary in order for the
// * results of one rule in a sequential program to be visible to the
// * next rule in that sequential program.
// */
//
// taskFactory = new RunRuleAndFlushBufferTaskFactory(taskFactory);
//
// }
return taskFactory;
}
final public IEvaluationPlanFactory getPlanFactory() {
return planFactory;
}
final public IResourceLocator getRelationLocator() {
return resourceLocator;
}
final public IBuffer newUnsynchronizedBuffer(
final IBuffer targetBuffer, final int chunkCapacity) {
// MAY be null.
final IElementFilter filter = getSolutionFilter();
return new UnsynchronizedArrayBuffer(targetBuffer,
chunkCapacity, ISolution.class, filter);
}
/**
* Note: {@link ISolution} (not relation elements) will be written on the
* buffer concurrently by different rules so there is no natural order for
* the elements in the buffer.
*/
final public IBlockingBuffer newQueryBuffer() {
if (getAction().isMutation())
throw new IllegalStateException();
return new BlockingBuffer(chunkOfChunksCapacity,
chunkCapacity, chunkTimeout, chunkTimeoutUnit);
}
/**
* {@inheritDoc}
*
* Note: {@link #getSolutionFilter()} is applied by
* {@link #newUnsynchronizedBuffer(IBuffer, int)} and NOT by the buffer
* returned by this method.
*/
@SuppressWarnings("unchecked")
public IBuffer newInsertBuffer(final IMutableRelation relation) {
if (getAction() != ActionEnum.Insert)
throw new IllegalStateException();
if (log.isDebugEnabled()) {
log.debug("relation=" + relation);
}
/*
* Buffer resolves the computed elements and writes them on the
* statement indices.
*/
return new AbstractSolutionBuffer.InsertSolutionBuffer(
chunkOfChunksCapacity, relation);
}
/**
* {@inheritDoc}
*
* Note: {@link #getSolutionFilter()} is applied by
* {@link #newUnsynchronizedBuffer(IBuffer, int)} and NOT by the buffer
* returned by this method.
*/
@SuppressWarnings("unchecked")
public IBuffer newDeleteBuffer(final IMutableRelation relation) {
if (getAction() != ActionEnum.Delete)
throw new IllegalStateException();
if (log.isDebugEnabled()) {
log.debug("relation=" + relation);
}
return new AbstractSolutionBuffer.DeleteSolutionBuffer(
chunkOfChunksCapacity, relation);
}
/**
* Return the {@link ISortKeyBuilder} used to impose DISTINCT on the
* solutions generated by a query.
*
* @param head
* The head of the rule.
*
* @return The {@link ISortKeyBuilder}.
*
* @todo This should be based on bop annotations and a hash table for
* distinct unless it is very high volume and you can wait for the
* first result, in which case a SORT should be selected. For high
* volume with low latency to the first result, use a persistent hash
* table on a temporary store.
*/
abstract protected ISortKeyBuilder newSortKeyBuilder(
final IPredicate head);
@SuppressWarnings("unchecked")
public IChunkedOrderedIterator runQuery(final IStep step)
throws Exception {
if (step == null)
throw new IllegalArgumentException();
if(log.isInfoEnabled())
log.info("program="+step.getName());
if(isEmptyProgram(step)) {
log.warn("Empty program");
return (IChunkedOrderedIterator) new EmptyProgramTask(
ActionEnum.Query, step).call();
}
final IChunkedOrderedIterator itr = (IChunkedOrderedIterator) runProgram(
ActionEnum.Query, step);
if (step.isRule() && ((IRule) step).getQueryOptions().isDistinct()) {
/*
* Impose a DISTINCT constraint based on the variable bindings
* selected by the head of the rule. The DistinctFilter will be
* backed by a TemporaryStore if more than one chunk of solutions is
* generated. That TemporaryStore will exist on the client where
* this method (runQuery) was executed. The TemporaryStore will be
* finalized and deleted when it is no longer referenced.
*/
final ISortKeyBuilder sortKeyBuilder;
if (((IRule) step).getHead() != null
&& (solutionFlags & ELEMENT) != 0) {
/*
* Head exists and elements are requested, so impose DISTINCT
* based on the materialized elements.
*/
sortKeyBuilder = new DelegateSortKeyBuilder(
newSortKeyBuilder(((IRule) step).getHead())) {
protected Object resolve(Object solution) {
return ((ISolution) solution).get();
}
};
} else {
if ((solutionFlags & BINDINGS) != 0) {
/*
* Bindings were requested so impose DISTINCT based on those
* bindings.
*/
sortKeyBuilder = new DelegateSortKeyBuilder(
newBindingSetSortKeyBuilder((IRule) step)) {
protected IBindingSet resolve(ISolution solution) {
return solution.getBindingSet();
}
};
} else {
throw new UnsupportedOperationException(
"You must specify BINDINGS since the rule does not have a head: "
+ step);
}
}
return new ChunkedConvertingIterator(itr,
new DistinctFilter(indexManager) {
protected byte[] getSortKey(ISolution e) {
return sortKeyBuilder.getSortKey(e);
}
});
}
return itr;
}
final public long runMutation(final IStep step) throws Exception {
if (step == null)
throw new IllegalArgumentException();
if (!action.isMutation())
throw new IllegalStateException();
if (step.isRule() && ((IRule) step).getHead() == null) {
throw new IllegalArgumentException("No head for this rule: " + step);
}
if(log.isInfoEnabled())
log.info("action=" + action + ", program=" + step.getName());
if(isEmptyProgram(step)) {
log.warn("Empty program");
return (Long) new EmptyProgramTask(action, step).call();
}
return (Long) runProgram(action, step);
}
/**
* Return true iff the step is an empty {@link IProgram}.
*
* @param step
* The step.
*/
final protected boolean isEmptyProgram(final IStep step) {
if (!step.isRule() && ((IProgram) step).stepCount() == 0) {
return true;
}
return false;
}
/**
* Core impl. This handles the logic required to execute the program either
* on a target {@link DataService} (highly efficient) or within the client
* using the {@link IClientIndex} to submit operations to the appropriate
* {@link DataService}(s) (not very efficient, even w/o RMI).
*
* @return Either an {@link IChunkedOrderedIterator} (query) or {@link Long}
* (mutation count).
*/
final protected Object runProgram(final ActionEnum action, final IStep step)
throws Exception {
if (action == null)
throw new IllegalArgumentException();
if (step == null)
throw new IllegalArgumentException();
final IIndexManager indexManager = getIndexManager();
if (indexManager instanceof IBigdataFederation) {
// distributed program execution.
return runDistributedProgram((IBigdataFederation) indexManager,
action, step);
} else {
// local Journal or TemporaryStore execution.
return runLocalProgram(action, step);
}
}
/**
* This variant handles both local indices on a {@link TemporaryStore} or
* {@link Journal} WITHOUT concurrency controls (fast).
*/
final protected Object runLocalProgram(final ActionEnum action,
final IStep step) throws Exception {
if (log.isInfoEnabled())
log.info("Running local program: action=" + action + ", program="
+ step.getName());
final IProgramTask innerTask = new ProgramTask(action, step,
getJoinNexusFactory(), getIndexManager());
return innerTask.call();
}
/**
* Runs a distributed {@link IProgram} (key-range partitioned indices, RMI,
* and multi-machine).
*/
final protected Object runDistributedProgram(final IBigdataFederation fed,
final ActionEnum action, final IStep step) throws Exception {
if (log.isInfoEnabled()) {
log.info("Running distributed program: action=" + action
+ ", program=" + step.getName());
}
final IProgramTask innerTask = new ProgramTask(action, step,
getJoinNexusFactory(), getIndexManager());
return innerTask.call();
}
// /**
// * This variant is submitted and executes the rules from inside of the
// * {@link ConcurrencyManager} on the {@link LocalDataServiceImpl} (fast).
// *
// * Note: This can only be done if all indices for the relation(s) are (a)
// * unpartitioned; and (b) located on the SAME {@link DataService}. This is
// * true for {@link LocalDataServiceFederation}. All other
// * {@link IBigdataFederation} implementations are scale-out (use key-range
// * partitioned indices).
// */
// protected Object runDataServiceProgram(final DataService dataService,
// final ActionEnum action, final IStep step)
// throws InterruptedException, ExecutionException {
//
// if (log.isInfoEnabled()) {
//
// log.info("Submitting program to data service: action=" + action
// + ", program=" + step.getName() + ", dataService="
// + dataService);
//
// }
//
// final IProgramTask innerTask = new ProgramTask(action, step,
// getJoinNexusFactory());
//
// return dataService.submit(innerTask).get();
//
// }
// /**
// * Return true if the relationName is on a
// * {@link TempTripleStore}
// *
// * @todo Rather than parsing the relation name, it would be better to have
// * the temporary store UUIDs explicitly declared.
// */
// protected boolean isTempStore(String relationName) {
//
// /* This is a typical UUID-based temporary store relation name.
// *
// * 1 2 3
// * 01234567890123456789012345678901234567
// * 81ad63b9-2172-45dc-bd97-03b63dfe0ba0kb.spo
// */
//
// if (relationName.length() > 37) {
//
// /*
// * Could be a relation on a temporary store.
// */
// if ( relationName.charAt( 8) == '-' //
// && relationName.charAt(13) == '-' //
// && relationName.charAt(18) == '-' //
// && relationName.charAt(23) == '-' //
// && relationName.charAt(38) == '.' //
// ) {
//
// /*
// * Pretty certain to be a relation on a temporary store.
// */
//
// return true;
//
// }
//
// }
//
// return false;
//
// }
//// /**
//// * A per-relation reentrant read-write lock allows either concurrent readers
//// * or an writer on the unisolated view of a relation. When we use this lock
//// * we also use {@link ITx#UNISOLATED} reads and writes and
//// * {@link #makeWriteSetsVisible()} is a NOP.
//// */
//// final private static boolean useReentrantReadWriteLockAndUnisolatedReads = true;
//
// public long getReadTimestamp(String relationName) {
//
//// if (useReentrantReadWriteLockAndUnisolatedReads) {
//
//// if (action.isMutation()) {
////
//// assert readTimestamp == ITx.UNISOLATED : "readTimestamp="+readTimestamp;
////
//// }
//
// return readTimestamp;
//
//// } else {
////
//// /*
//// * When the relation is the focusStore choose {@link ITx#UNISOLATED}.
//// * Otherwise choose whatever was specified to the
//// * {@link RDFJoinNexusFactory}. This is because we avoid doing a
//// * commit on the focusStore and instead just its its UNISOLATED
//// * indices. This is more efficient since they are already buffered
//// * and since we can avoid touching disk at all for small data sets.
//// */
////
//// if (isTempStore(relationName)) {
////
//// return ITx.UNISOLATED;
////
//// }
////
//// if (lastCommitTime != 0L && action.isMutation()) {
////
//// /*
//// * Note: This advances the read-behind timestamp for a local
//// * Journal configuration without the ConcurrencyManager (the
//// * only scenario where we do an explicit commit).
//// */
////
//// return TimestampUtility.asHistoricalRead(lastCommitTime);
////
//// }
////
//// return readTimestamp;
////
//// }
//
// }
}