Various aspects of OptaPlanner configuration can be specified in the XML config then overridden via JavaAPI. For example you can set a default TerminationConfig in your XML then at runtime switch it for a user specified value.
The docs seem to indicate that it should be possible, but I haven't found any examples. The context is that I'm hoping to dynamically alter the UnionMove section of the config and omit Swap moves if the instance to be solved doesn't have at least 2 planning entities. Currently this results in lots of Bailing from infinite logging when users are experimenting with small problems.
Taken from https://www.optaplanner.org/docs/optaplanner/latest/planner-configuration/planner-configuration.html#solverConfigurationByJavaAPI :
private SolverConfig template;
public void init() {
template = SolverConfig.createFromXmlResource(
"org/optaplanner/examples/nqueens/solver/nqueensSolverConfig.xml");
template.setTerminationConfig(new TerminationConfig());
}
// Called concurrently from different threads
public void userRequest(..., long userInput) {
SolverConfig solverConfig = new SolverConfig(template); // Copy it
solverConfig.getTerminationConfig().setMinutesSpentLimit(userInput);
SolverFactory<NQueens> solverFactory = SolverFactory.create(solverConfig);
Solver<NQueens> solver = solverFactory.buildSolver();
...
}
In your case, you would need to modify the Phase List of the SolverConfig (see the SolverConfig Javadocs ); the move selectors can be gotten and set via LocalSearchPhaseConfig and ConstructionHeuristicPhaseConfig (you need to use instanceof when iterating phase list to determine which one to cast to).
The bailing from infinite logging sound like a bug.
Related
I am building a Capacited Vehicle Routing Problem with Time Windows, but with one small difference when compared to the one provided in examples from the documentation: I don't have a depot. Instead, each order has a pickup step, and a delivery step, in two different locations.
(like in the Vehicle Routing example from the documentation, the previousStep planning variable has the CHAINED graph type, and its valueRangeProviderRefs includes both Drivers, and Steps)
This difference adds a couple of constraints:
the pickup and delivery steps of a given order must be handled by the same driver
the pickup must be before the delivery
After experimenting with constraints, I have found that it would be more efficient to implement two types of custom moves:
assign both steps of an order to a driver
rearrange the steps of a driver
I am currently implementing that first custom move. My solver's configuration looks like this:
SolverFactory<RoutingProblem> solverFactory = SolverFactory.create(
new SolverConfig()
.withSolutionClass(RoutingProblem.class)
.withEntityClasses(Step.class, StepList.class)
.withScoreDirectorFactory(new ScoreDirectorFactoryConfig()
.withConstraintProviderClass(Constraints.class)
)
.withTerminationConfig(new TerminationConfig()
.withSecondsSpentLimit(60L)
)
.withPhaseList(List.of(
new LocalSearchPhaseConfig()
.withMoveSelectorConfig(CustomMoveListFactory.getConfig())
))
);
My CustomMoveListFactory looks like this (I plan on migrating it to an MoveIteratorFactory later, but for the moment, this is easier to read and write):
public class CustomMoveListFactory implements MoveListFactory<RoutingProblem> {
public static MoveListFactoryConfig getConfig() {
MoveListFactoryConfig result = new MoveListFactoryConfig();
result.setMoveListFactoryClass(CustomMoveListFactory.class);
return result;
}
#Override
public List<? extends Move<RoutingProblem>> createMoveList(RoutingProblem routingProblem) {
List<Move<RoutingProblem>> moves = new ArrayList<>();
// 1. Assign moves
for (Order order : routingProblem.getOrders()) {
Driver currentDriver = order.getDriver();
for (Driver driver : routingProblem.getDrivers()) {
if (!driver.equals(currentDriver)) {
moves.add(new AssignMove(order, driver));
}
}
}
// 2. Rearrange moves
// TODO
return moves;
}
}
And finally, the move itself looks like this (nevermind the undo or the isDoable for the moment):
#Override
protected void doMoveOnGenuineVariables(ScoreDirector<RoutingProblem> scoreDirector) {
assignStep(scoreDirector, order.getPickupStep());
assignStep(scoreDirector, order.getDeliveryStep());
}
private void assignStep(ScoreDirector<RoutingProblem> scoreDirector, Step step) {
StepList beforeStep = step.getPreviousStep();
Step afterStep = step.getNextStep();
// 1. Insert step at the end of the driver's step list
StepList lastStep = driver.getLastStep();
scoreDirector.beforeVariableChanged(step, "previousStep"); // NullPointerException here
step.setPreviousStep(lastStep);
scoreDirector.afterVariableChanged(step, "previousStep");
// 2. Remove step from current chained list
if (afterStep != null) {
scoreDirector.beforeVariableChanged(afterStep, "previousStep");
afterStep.setPreviousStep(beforeStep);
scoreDirector.afterVariableChanged(afterStep, "previousStep");
}
}
The idea being that at no point I'm doing an invalid chained list manipulation:
However, as the title and the code comment indicate, I am getting a NullPointerException when I call scoreDirector.beforeVariableChanged. None of my variables are null (I've printed them to make sure). The NullPointerException doesn't occur in my code, but deep inside Optaplanner's inner workings, making it difficult for me to fix it:
Exception in thread "main" java.lang.NullPointerException
at org.drools.core.common.NamedEntryPoint.update(NamedEntryPoint.java:353)
at org.drools.core.common.NamedEntryPoint.update(NamedEntryPoint.java:338)
at org.drools.core.impl.StatefulKnowledgeSessionImpl.update(StatefulKnowledgeSessionImpl.java:1579)
at org.drools.core.impl.StatefulKnowledgeSessionImpl.update(StatefulKnowledgeSessionImpl.java:1551)
at org.optaplanner.core.impl.score.stream.drools.DroolsConstraintSession.update(DroolsConstraintSession.java:49)
at org.optaplanner.core.impl.score.director.stream.ConstraintStreamScoreDirector.afterVariableChanged(ConstraintStreamScoreDirector.java:137)
at org.optaplanner.core.impl.domain.variable.inverserelation.SingletonInverseVariableListener.retract(SingletonInverseVariableListener.java:96)
at org.optaplanner.core.impl.domain.variable.inverserelation.SingletonInverseVariableListener.beforeVariableChanged(SingletonInverseVariableListener.java:46)
at org.optaplanner.core.impl.domain.variable.listener.support.VariableListenerSupport.beforeVariableChanged(VariableListenerSupport.java:170)
at org.optaplanner.core.impl.score.director.AbstractScoreDirector.beforeVariableChanged(AbstractScoreDirector.java:430)
at org.optaplanner.core.impl.score.director.AbstractScoreDirector.beforeVariableChanged(AbstractScoreDirector.java:390)
at test.optaplanner.solver.AssignMove.assignStep(AssignMove.java:98)
at test.optaplanner.solver.AssignMove.doMoveOnGenuineVariables(AssignMove.java:85)
at org.optaplanner.core.impl.heuristic.move.AbstractMove.doMove(AbstractMove.java:35)
at org.optaplanner.core.impl.heuristic.move.AbstractMove.doMove(AbstractMove.java:30)
at org.optaplanner.core.impl.score.director.AbstractScoreDirector.doAndProcessMove(AbstractScoreDirector.java:187)
at org.optaplanner.core.impl.localsearch.decider.LocalSearchDecider.doMove(LocalSearchDecider.java:132)
at org.optaplanner.core.impl.localsearch.decider.LocalSearchDecider.decideNextStep(LocalSearchDecider.java:116)
at org.optaplanner.core.impl.localsearch.DefaultLocalSearchPhase.solve(DefaultLocalSearchPhase.java:70)
at org.optaplanner.core.impl.solver.AbstractSolver.runPhases(AbstractSolver.java:98)
at org.optaplanner.core.impl.solver.DefaultSolver.solve(DefaultSolver.java:189)
at test.optaplanner.OptaPlannerService.testOptaplanner(OptaPlannerService.java:68)
at test.optaplanner.App.main(App.java:13)
Is there something I did wrong? It seems I am following the documentation for custom moves fairly closely, outside of the fact that I am using exclusively java code instead of drools.
The initial solution I feed to the solver has all of the steps assigned to a single driver. There are 15 drivers and 40 orders.
In order to bypass this error, I have tried a number of different things:
remove the shadow variable annotation, turn Driver into a problem fact, and handle the nextStep field myself => this makes no difference
use Simulated Annealing + First Fit Decreasing construction heuristics, and start with steps not assigned to any driver (this was inspired by looking up the example here, which is more complete than the one from the documentation) => the NullPointerException appears on afterVariableChanged instead, but it still appears.
a number of other things which were probably not very smart
But without a more helpful error message, I can't think of anything else to try.
Thank you for your help
I wish to access many classes and variables, I would like to do this by dynamically setting the class name and variable name. Currently I am using
MyClass["myVariable1"]
to dynamically access the variable name
MyClass.myVariable1
I want to also dynanmically acces the class name, something like
["MyClass"]["myVariable1"]
But this does not work.
The purpose is that I have shared object with many user settings, I want to iterate through the shared object and set all the user settings across all the classes. I think if I cant dynamically access the class I must have a statement for each and every class name/variable.
I advise against such a practice. Although technically possible, it is like welcoming a disaster into the app architecture:
You rely on something you have no apparent control of: on the way Flash names the classes.
You walk out of future possibility to protect your code with identifier renaming obfuscation because it will render your code invalid.
Compile time error checks is better than runtime, and you are leaving it to runtime. If it happens to fail in non-debug environment, you will never know.
The next developer to work with your code (might be you in a couple of years) will have hard time finding where the initial data coming from.
So, having all of above, I encourage you to switch to another model:
package
{
import flash.net.SharedObject;
public class SharedData
{
static private var SO:SharedObject;
static public function init():void
{
SO = SharedObject.getLocal("my_precious_shared_data", "/");
}
static public function read(key:String):*
{
// if (!SO) init();
return SO.data[key];
}
static public function write(key:String, value:*):void
{
// if (!SO) init();
SO.data[key] = value;
SO.flush();
}
// Returns stored data if any, or default value otherwise.
// A good practice of default application values that might
// change upon user activity, e.g. sound volume or level progress.
static public function readSafe(key:String, defaultValue:*):*
{
// if (!SO) init();
return SO.data.hasOwnProperty(key)? read(key): defaultValue;
}
}
}
In the main class you call
SharedData.init();
// So now your shared data are available.
// If you are not sure you can call it before other classes will read
// the shared data, just uncomment // if (!SO) init(); lines in SharedData methods.
Then each class that feeds on these data should have an initialization block:
// It's a good idea to keep keys as constants
// so you won't occasionally mistype them.
// Compile time > runtime again.
static private const SOMAXMANA:String = "maxmana";
static private const SOMAXHP:String = "maxhp";
private var firstTime:Boolean = true;
private var maxmana:int;
private var maxhp:int;
// ...
if (firstTime)
{
// Make sure it does not read them second time.
firstTime = false;
maxhp = SharedData.readSafe(SOMAXHP, 100);
maxmana = SharedData.readSafe(SOMAXMANA, 50);
}
Well, again. The code above:
does not employ weird practices and easy to understand
in each class anyone can clearly see where the data come from
will be checked for errors at compile time
can be obfuscated and protected
You can try getting the class into a variable and going from there:
var myClass:Class = getDefinitionByName("MyClass") as Class;
myClass["myVariable1"] = x;
I want to use something like Cucumber JVM to drive performance tests written for Gatling.
Ideally the Cucumber features would somehow build a scenario dynamically - probably reusing predefined chain objects similar to the method described in the "Advanced Tutorial", e.g.
val scn = scenario("Scenario Name").exec(Search.search("foo"), Browse.browse, Edit.edit("foo", "bar")
I've looked at how the Maven plugin executes the scripts, and I've also seen mention of using an App trait but I can't find any documentation for the later and it strikes me that somebody else will have wanted to do this before...
Can anybody point (a Gatling noob) in the direction of some documentation or example code of how to achieve this?
EDIT 20150515
So to explain a little more:
I have created a trait which is intended to build up a sequence of, I think, ChainBuilders that are triggered by Cucumber steps:
trait GatlingDsl extends ScalaDsl with EN {
private val gatlingActions = new ArrayBuffer[GatlingBehaviour]
def withGatling(action: GatlingBehaviour): Unit = {
gatlingActions += action
}
}
A GatlingBehaviour would look something like:
object Google {
class Home extends GatlingBehaviour {
def execute: ChainBuilder =
exec(http("Google Home")
.get("/")
)
}
class Search extends GatlingBehaviour {...}
class FindResult extends GatlingBehaviour {...}
}
And inside the StepDef class:
class GoogleStepDefinitions extends GatlingDsl {
Given( """^the Google search page is displayed$""") { () =>
println("Loading www.google.com")
withGatling(Home())
}
When( """^I search for the term "(.*)"$""") { (searchTerm: String) =>
println("Searching for '" + searchTerm + "'...")
withGatling(Search(searchTerm))
}
Then( """^"(.*)" appears in the search results$""") { (expectedResult: String) =>
println("Found " + expectedResult)
withGatling(FindResult(expectedResult))
}
}
The idea being that I can then execute the whole sequence of actions via something like:
val scn = Scenario(cucumberScenario).exec(gatlingActions)
setup(scn.inject(atOnceUsers(1)).protocols(httpConf))
and then check the reports or catch an exception if the test fails, e.g. response time too long.
It seems that no matter how I use the 'exec' method it tries to instantly execute it there and then, not waiting for the scenario.
Also I don't know if this is the best approach to take, we'd like to build some reusable blocks for our Gatling tests that can be constructed via Cucumber's Given/When/Then style. Is there a better or already existing approach?
Sadly, it's not currently feasible to have Gatling directly start a Simulation instance.
Not that's it's not technically feasible, but you're just the first person to try to do this.
Currently, Gatling is usually in charge of compiling and can only be passed the name of the class to load, not an instance itself.
You can maybe start by forking io.gatling.app.Gatling and io.gatling.core.runner.Runner, and then provide a PR to support this new behavior. The former is the main entry point, and the latter the one can instanciate and run the simulation.
I recently ran into a similar situation, and did not want to fork gatling. And while this solved my immediate problem, it only partially solves what you are trying to do, but hopefully someone else will find this useful.
There is an alternative. Gatling is written in Java and Scala so you can call Gatling.main directly and pass it the arguments you need to run the Gatling Simulation you want. The problem is, the main explicitly calls System.exit so you have to also use a custom security manager to prevent it from actually exiting.
You need to know two things:
the class (with the full package) of the Simulation you want to run
example: com.package.your.Simulation1
the path where the binaries are compiled.
The code to run a Simulation:
protected void fire(String gatlingGun, String binaries){
SecurityManager sm = System.getSecurityManager();
System.setSecurityManager(new GatlingSecurityManager());
String[] args = {"--simulation", gatlingGun,
"--results-folder", "gatling-results",
"--binaries-folder", binaries};
try {
io.gatling.app.Gatling.main(args);
}catch(SecurityException se){
LOG.debug("gatling test finished.");
}
System.setSecurityManager(sm);
}
The simple security manager i used:
public class GatlingSecurityManager extends SecurityManager {
#Override
public void checkExit(int status){
throw new SecurityException("Tried to exit.");
}
#Override
public void checkPermission(Permission perm) {
return;
}
}
The problem is then getting the information you want out of the simulation after it has been run.
I want to add a tag #skiponchrome to a scenario, this should skip the scenario when running a Selenium test with the Chrome browser. The reason to-do this is because some scenario's work in some environments and not in others, this might not even be browser testing specific and could be applied in other situation for example OS platforms.
Example hook:
#Before("#skiponchrome") // this works
public void beforeScenario() {
if(currentBrowser == 'chrome') { // this works
// Skip scenario code here
}
}
I know it is possible to define ~#skiponchrome in the cucumber tags to skip the tag, but I would like to skip a tag at run-time. This way I don't have to think about which steps to skip in advance when I starting a test run on a certain environment.
I would like to create a hook that catches the tag and skips the scenario without reporting a fail/error. Is this possible?
I realized that this is a late update to an already answered question, but I want to add one more option directly supported by cucumber-jvm:
#Before //(cucumber one)
public void setup(){
Assume.assumeTrue(weAreInPreProductionEnvironment);
}
"and the scenario will be marked as ignored (but the test will pass) if weAreInPreProductionEnvironment is false."
You will need to add
import org.junit.Assume;
The major difference with the accepted answer is that JUnit assume failures behave just like pending
Important Because of a bug fix you will need cucumber-jvm release 1.2.5 which as of this writing is the latest. For example, the above will generate a failure instead of a pending in cucumber-java8-1.2.3.jar
I really prefer to be explicit about which tests are being run, by having separate run configurations defined for each environment. I also like to keep the number of tags I use to a minimum, to keep the number of configurations manageable.
I don't think it's possible to achieve what you want with tags alone. You would need to write a custom jUnit test runner to use in place of #RunWith(Cucumber.class). Take a look at the Cucumber implementation to see how things work. You would need to alter the RuntimeOptions created by the RuntimeOptionsFactory to include/exclude tags depending on the browser, or other runtime condition.
Alternatively, you could consider writing a small script which invokes your test suite, building up a list of tags to include/exclude dynamically, depending on the environment you're running in. I would consider this to be a more maintainable, cleaner solution.
It's actually really easy. If you dig though the Cucumber-JVM and JUnit 4 source code, you'll find that JUnit makes skipping during runtime very easy (just undocumented).
Take a look at the following source code for JUnit 4's ParentRunner, which Cucumber-JVM's FeatureRunner (which is used in Cucumber, the default Cucumber runner):
#Override
public void run(final RunNotifier notifier) {
EachTestNotifier testNotifier = new EachTestNotifier(notifier,
getDescription());
try {
Statement statement = classBlock(notifier);
statement.evaluate();
} catch (AssumptionViolatedException e) {
testNotifier.fireTestIgnored();
} catch (StoppedByUserException e) {
throw e;
} catch (Throwable e) {
testNotifier.addFailure(e);
}
}
This is how JUnit decides what result to show. If it's successful it will show a pass, but it's possible to #Ignore in JUnit, so what happens in that case? Well, an AssumptionViolatedException is thrown by the RunNotifier (or Cucumber FeatureRunner in this case).
So your example becomes:
#Before("#skiponchrome") // this works
public void beforeScenario() {
if(currentBrowser == 'chrome') { // this works
throw new AssumptionViolatedException("Not supported on Chrome")
}
}
If you've used vanilla JUnit 4 before, you'd remember that #Ignore takes an optional message that is displayed when a test is ignored by the runner. AssumptionViolatedException carries the message, so you should see it in your test output after a test is skipped this way without having to write your own custom runner.
I too had the same challenge, where in I need to skip a scenario from running based on a flag which I obtain from the application dynamically in run-time, which tells whether the feature to be tested is enabled on the application or not..
so this is how I wrote my logic in the scenarios file, where we have the glue code for each step.
I have used a unique tag '#Feature-01AXX' to mark my scenarios that need to be run only when that feature(code) is available on the application.
so for every scenario, the tag '#Feature-01XX' is checked first, if its present then the check for the availability of the feature is made, only then the scenario will be picked for running. Else it will be merely skipped, and Junit will not mark this as failure, instead it will me marked as Pass. So the final result if these tests did not run due to the un-availability of the feature will be pass, that's cool...
#Before
public void before(final Scenario scenario) throws Exception {
/*
my other pre-setup tasks for each scenario.
*/
// get all the scenario tags from the scenario head.
final ArrayList<String> scenarioTags = new ArrayList<>();
scenarioTags.addAll(scenario.getSourceTagNames());
// check if the feature is enabled on the appliance, so that the tests can be run.
if (checkForSkipScenario(scenarioTags)) {
throw new AssumptionViolatedException("The feature 'Feature-01AXX' is not enabled on this appliance, so skipping");
}
}
private boolean checkForSkipScenario(final ArrayList<String> scenarioTags) {
// I use a tag "#Feature-01AXX" on the scenarios which needs to be run when the feature is enabled on the appliance/application
if (scenarioTags.contains("#Feature-01AXX") && !isTheFeatureEnabled()) { // if feature is not enabled, then we need to skip the scenario.
return true;
}
return false;
}
private boolean isTheFeatureEnabled(){
/*
my logic to check if the feature is available/enabled on the application.
in my case its an REST api call, I parse the JSON and check if the feature is enabled.
if it is enabled return 'true', else return 'false'
*/
}
I've implemented a customized junit runner as below. The idea is to add tags during runtime.
So say for a scenario we need new users, we tag the scenarios as "#requires_new_user". Then if we run our test in an environment (say production environment which dose not allow you to register new user easily), then we will figure out that we are not able to get new user. Then the ""not #requires_new_user" will be added to cucumber options to skip the scenario.
This is the most clean solution I can imagine now.
public class WebuiCucumberRunner extends ParentRunner<FeatureRunner> {
private final JUnitReporter jUnitReporter;
private final List<FeatureRunner> children = new ArrayList<FeatureRunner>();
private final Runtime runtime;
private final Formatter formatter;
/**
* Constructor called by JUnit.
*
* #param clazz the class with the #RunWith annotation.
* #throws java.io.IOException if there is a problem
* #throws org.junit.runners.model.InitializationError if there is another problem
*/
public WebuiCucumberRunner(Class clazz) throws InitializationError, IOException {
super(clazz);
ClassLoader classLoader = clazz.getClassLoader();
Assertions.assertNoCucumberAnnotatedMethods(clazz);
RuntimeOptionsFactory runtimeOptionsFactory = new RuntimeOptionsFactory(clazz);
RuntimeOptions runtimeOptions = runtimeOptionsFactory.create();
addTagFiltersAsPerTestRuntimeEnvironment(runtimeOptions);
ResourceLoader resourceLoader = new MultiLoader(classLoader);
runtime = createRuntime(resourceLoader, classLoader, runtimeOptions);
formatter = runtimeOptions.formatter(classLoader);
final JUnitOptions junitOptions = new JUnitOptions(runtimeOptions.getJunitOptions());
final List<CucumberFeature> cucumberFeatures = runtimeOptions.cucumberFeatures(resourceLoader, runtime.getEventBus());
jUnitReporter = new JUnitReporter(runtime.getEventBus(), runtimeOptions.isStrict(), junitOptions);
addChildren(cucumberFeatures);
}
private void addTagFiltersAsPerTestRuntimeEnvironment(RuntimeOptions runtimeOptions)
{
String channel = Configuration.TENANT_NAME.getValue().toLowerCase();
runtimeOptions.getTagFilters().add("#" + channel);
if (!TestEnvironment.getEnvironment().isNewUserAvailable()) {
runtimeOptions.getTagFilters().add("not #requires_new_user");
}
}
...
}
Or you can extends the official Cucumber Junit test runner cucumber.api.junit.Cucumber and override method
/**
* Create the Runtime. Can be overridden to customize the runtime or backend.
*
* #param resourceLoader used to load resources
* #param classLoader used to load classes
* #param runtimeOptions configuration
* #return a new runtime
* #throws InitializationError if a JUnit error occurred
* #throws IOException if a class or resource could not be loaded
* #deprecated Neither the runtime nor the backend or any of the classes involved in their construction are part of
* the public API. As such they should not be exposed. The recommended way to observe the cucumber process is to
* listen to events by using a plugin. For example the JSONFormatter.
*/
#Deprecated
protected Runtime createRuntime(ResourceLoader resourceLoader, ClassLoader classLoader,
RuntimeOptions runtimeOptions) throws InitializationError, IOException {
ClassFinder classFinder = new ResourceLoaderClassFinder(resourceLoader, classLoader);
return new Runtime(resourceLoader, classFinder, classLoader, runtimeOptions);
}
You can manipulate runtimeOptions here as you wish. But the method is marked as deprecated, so use it with caution.
If you're using Maven, you could read use a browser profile and then set the appropriate ~ exclude tags there?
Unless you're asking how to run this from command line, in which case you tag the scenario with #skipchrome and then when you run cucumber set the cucumber options to tags = {"~#skipchrome"}
If you wish simply to temporarily skip a scenario (for example, while writing the scenarios), you can comment it out (ctrl+/ in Eclipse or Intellij).
I've recently started working on a new application which will utilize task parallelism. I have just begun writing a tasking framework, but have recently seen a number of posts on SO regarding the new System.Threading.Tasks namespace which may be useful to me (and I would rather use an existing framework than roll my own).
However looking over MSDN I haven't seen how / if, I can implement the functionality which I'm looking for:
Dependency on other tasks completing.
Able to wait on an unknown number of tasks preforming the same action (maybe wrapped in the same task object which is invoked multiple times)
Set maximum concurrent instances of a task since they use a shared resource there is no point running more than one at once
Hint at priority, or scheduler places tasks with lower maximum concurrent instances at a higher priority (so as to keep said resource in use as much as possible)
Edit ability to vary the priority of tasks which are preforming the same action (pretty poor example but, PredictWeather (Tommorrow) will have a higher priority than PredictWeather (NextWeek))
Can someone point me towards an example / tell me how I can achieve this? Cheers.
C# Use Case: (typed in SO so please for give any syntax errors / typos)
**note Do() / DoAfter() shouldn't block the calling thread*
class Application ()
{
Task LeafTask = new Task (LeafWork) {PriorityHint = High, MaxConcurrent = 1};
var Tree = new TaskTree (LeafTask);
Task TraverseTask = new Task (Tree.Traverse);
Task WorkTask = new Task (MoreWork);
Task RunTask = new Task (Run);
Object SharedLeafWorkObject = new Object ();
void Entry ()
{
RunTask.Do ();
RunTask.Join (); // Use this thread for task processing until all invocations of RunTask are complete
}
void Run(){
TraverseTask.Do ();
// Wait for TraverseTask to make sure all leaf tasks are invoked before waiting on them
WorkTask.DoAfter (new [] {TraverseTask, LeafTask});
if (running){
RunTask.DoAfter (WorkTask); // Keep at least one RunTask alive to prevent Join from 'unblocking'
}
else
{
TraverseTask.Join();
WorkTask.Join ();
}
}
void LeafWork (Object leaf){
lock (SharedLeafWorkObject) // Fake a shared resource
{
Thread.Sleep (200); // 'work'
}
}
void MoreWork ()
{
Thread.Sleep (2000); // this one takes a while
}
}
class TaskTreeNode<TItem>
{
Task LeafTask; // = Application::LeafTask
TItem Item;
void Traverse ()
{
if (isLeaf)
{
// LeafTask set in C-Tor or elsewhere
LeafTask.Do(this.Item);
//Edit
//LeafTask.Do(this.Item, this.Depth); // Deeper items get higher priority
return;
}
foreach (var child in this.children)
{
child.Traverse ();
}
}
}
There are numerous examples here:
http://code.msdn.microsoft.com/ParExtSamples
There's a great white paper which covers a lot of the details you mention above here:
"Patterns for Parallel Programming: Understanding and Applying Parallel Patterns with the .NET Framework 4"
http://www.microsoft.com/downloads/details.aspx?FamilyID=86b3d32b-ad26-4bb8-a3ae-c1637026c3ee&displaylang=en
Off the top of my head I think you can do all the things you list in your question.
Dependencies etc: Task.WaitAll(Task[] tasks)
Scheduler: The library supports numerous options for limiting number of threads in use and supports providing your own scheduler. I would avoid altering the priority of threads if at all possible. This is likely to have negative impact on the scheduler, unless you provide your own.