|Modifier||Constructor and Description|
|Modifier and Type||Method and Description|
Run the task.
Flush the buffer(s) and aggregate the mutation count from each buffer.
This just reads off and aggregates the mutationCount from each buffer as reported by
Builds a set of tasks for the program.
clone, getIndexNames, getMutationBuffers, getReadRelations, getWriteRelationNames, getWriteRelations, runOne, runParallel, runSequential, submit, toString
getDataService, isDataService, setDataService
equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait
Note: We can create the individual tasks that we need to execute now that we are in the correct execution context.
Note: The mutation tasks write on
IBuffers and those buffers
flush to indices in the
IMutableRelations. We have to defer the
creation of those buffers until we are in the execution context and have
access to the correct indices. In turn, this means that we can not create
the tasks that we are going to execute until we have those buffers on
hand. Hence everything gets deferred until we are in the correct
execution context and have the actual
IIndexManager with which
the tasks will execute.
protected void flushBuffers(IJoinNexus joinNexus, Map<String,IBuffer<ISolution>> buffers) throws InterruptedException, ExecutionException
protected long getMutationCountFromBuffers(RuleStats totals, Map<String,IBuffer<ISolution>> buffers)
IBuffer.flush(). This is the actual mutation count for the step(s) executed by the
Note: The buffers SHOULD already have been flushed as this does NOT
parallelise the writes on the
#flushBuffers(IJoinNexus, RuleStats, Map), which does
parallelize those writes.
Copyright © 2006–2019 SYSTAP, LLC DBA Blazegraph. All rights reserved.