/// <summary>
/// Executes the Mutex benchmark with a problem size of n.
/// This variant uses IAction.Apply and IAction.MinMax to apply the rules
/// (also available IAction.ApplyStar and IAction.ApplyPlus).
/// These functions make it very easy to program complex rule applications,
/// while providing the same speed as using IAction.Match and IAction.Modify
/// (this example is not a complex scenario).
/// </summary>
/// <param name="n">The problem size, i.e. the number of Process nodes</param>
static void AlternativeTwo(int n)
{
int startTime = Environment.TickCount;
MutexGraph graph = new MutexGraph();
MutexPimpedActions actions = new MutexPimpedActions(graph);
LGSPGraphProcessingEnvironment procEnv = new LGSPGraphProcessingEnvironment(graph, actions);
LGSPNode p1 = graph.CreateNodeProcess();
LGSPNode p2 = graph.CreateNodeProcess();
LGSPEdge n1 = graph.CreateEdgenext(p1, p2);
LGSPEdge n2 = graph.CreateEdgenext(p2, p1);
Action_newRule.Instance.ApplyMinMax(procEnv, n - 2, n - 2);
Action_mountRule.Instance.Apply(procEnv);
Action_requestRule.Instance.ApplyMinMax(procEnv, n, n);
Action_takeRule takeRule = Action_takeRule.Instance;
Action_releaseRule releaseRule = Action_releaseRule.Instance;
Action_giveRule giveRule = Action_giveRule.Instance;
for (int i = 0; i < n; i++)
{
takeRule.Apply(procEnv);
releaseRule.Apply(procEnv);
giveRule.Apply(procEnv);
}
int endTime = Environment.TickCount;
PrintResults(endTime - startTime, graph);
}
/// <summary>
/// Executes the Mutex benchmark with a problem size of n.
/// This variant uses BaseActions.ApplyGraphRewriteSequence.
/// Although this provides the fastest way to program rule applications
/// for simple scenarios, it needs aprox. 12% more running time than
/// via Apply or Match/Modify.
/// </summary>
/// <param name="n">The problem size, i.e. the number of Process nodes</param>
static void AlternativeThree(int n)
{
int startTime = Environment.TickCount;
MutexGraph graph = new MutexGraph();
MutexPimpedActions actions = new MutexPimpedActions(graph);
LGSPGraphProcessingEnvironment procEnv = new LGSPGraphProcessingEnvironment(graph, actions);
LGSPNode p1 = graph.CreateNodeProcess();
LGSPNode p2 = graph.CreateNodeProcess();
LGSPEdge n1 = graph.CreateEdgenext(p1, p2);
LGSPEdge n2 = graph.CreateEdgenext(p2, p1);
procEnv.ApplyGraphRewriteSequence("newRule[" + (n - 2) + "] && mountRule && requestRule[" + n
+ "] && (takeRule && releaseRule && giveRule)[" + n + "]");
int endTime = Environment.TickCount;
PrintResults(endTime - startTime, graph);
}
/// <summary>
/// Executes the Mutex benchmark with a problem size of n.
/// This variant uses IAction.Match and IAction.Modify to apply the rules.
/// This way you get access to the found matches and can select between them manually.
/// </summary>
/// <param name="n">The problem size, i.e. the number of Process nodes</param>
static void AlternativeOne(int n)
{
int startTime = Environment.TickCount;
MutexGraph graph = new MutexGraph();
MutexPimpedActions actions = new MutexPimpedActions(graph);
LGSPGraphProcessingEnvironment procEnv = new LGSPGraphProcessingEnvironment(graph, actions);
LGSPNode p1 = graph.CreateNodeProcess();
LGSPNode p2 = graph.CreateNodeProcess();
LGSPEdge n1 = graph.CreateEdgenext(p1, p2);
LGSPEdge n2 = graph.CreateEdgenext(p2, p1);
IMatches matches;
Action_newRule newRule = Action_newRule.Instance;
for (int i = 0; i < n - 2; i++)
{
matches = newRule.Match(procEnv, 1);
newRule.Modify(procEnv, matches.First);
}
matches = Action_mountRule.Instance.Match(procEnv, 1);
Action_mountRule.Instance.Modify(procEnv, matches.First);
Action_requestRule requestRule = Action_requestRule.Instance;
for (int i = 0; i < n; i++)
{
matches = requestRule.Match(procEnv, 1);
requestRule.Modify(procEnv, matches.First);
}
/**
* The MutexPimped.grg file has been annotated with [prio] to generate very good static
* searchplans, without the need of dynamically generating searchplans.
*
* Otherwise you would use the following code to get better matcher programs:
*
* graph.AnalyzeGraph();
* LGSPAction[] newActions = actions.GenerateSearchPlans("takeRule", "releaseRule", "giveRule");
* LGSPAction takeRule = newActions[0];
* LGSPAction releaseRule = newActions[1];
* LGSPAction giveRule = newActions[2];
*
* Alternatively you could also write:
*
* graph.AnalyzeGraph();
* actions.GenerateSearchPlans("takeRule", "releaseRule", "giveRule");
* LGSPAction takeRule = actions.GetAction("takeRule");
* LGSPAction releaseRule = actions.GetAction("releaseRule");
* LGSPAction giveRule = actions.GetAction("giveRule");
*/
Action_takeRule takeRule = Action_takeRule.Instance;
Action_releaseRule releaseRule = Action_releaseRule.Instance;
Action_giveRule giveRule = Action_giveRule.Instance;
for (int i = 0; i < n; i++)
{
matches = takeRule.Match(procEnv, 1);
takeRule.Modify(procEnv, matches.First);
matches = releaseRule.Match(procEnv, 1);
releaseRule.Modify(procEnv, matches.First);
matches = giveRule.Match(procEnv, 1);
giveRule.Modify(procEnv, matches.First);
}
int endTime = Environment.TickCount;
PrintResults(endTime - startTime, graph);
// example for reading annotations at API level
Action_annotationTestRule annotationTestRule = Action_annotationTestRule.Instance;
foreach (KeyValuePair <string, string> annotation in annotationTestRule.rulePattern.Annotations)
{
Console.WriteLine("The rule " + annotationTestRule.Name + " is decorated with an annotation " + annotation.Key + " -> " + annotation.Value);
}
foreach (IPatternNode node in annotationTestRule.RulePattern.PatternGraph.Nodes)
{
foreach (KeyValuePair <string, string> annotation in node.Annotations)
{
Console.WriteLine("The pattern node " + node.Name + " of rule " + annotationTestRule.Name + " is decorated with an annotation " + annotation.Key + " -> " + annotation.Value);
}
}
IAnnotationTestNode testNode = graph.CreateNodeAnnotationTestNode();
foreach (KeyValuePair <string, string> annotation in testNode.Type.Annotations)
{
Console.WriteLine("The node type " + testNode.Type.Name + " is decorated with an annotation " + annotation.Key + " -> " + annotation.Value);
}
}
/// <summary>
/// Executes the Mutex benchmark with a problem size of n.
/// This variant uses IAction.Match and IAction.Modify to apply the rules.
/// This way you get access to the found matches and can select between them manually.
/// </summary>
/// <param name="n">The problem size, i.e. the number of Process nodes</param>
static void AlternativeOne(int n)
{
int startTime = Environment.TickCount;
MutexGraph graph = new MutexGraph();
MutexPimpedActions actions = new MutexPimpedActions(graph);
LGSPGraphProcessingEnvironment procEnv = new LGSPGraphProcessingEnvironment(graph, actions);
LGSPNode p1 = graph.CreateNodeProcess();
LGSPNode p2 = graph.CreateNodeProcess();
LGSPEdge n1 = graph.CreateEdgenext(p1, p2);
LGSPEdge n2 = graph.CreateEdgenext(p2, p1);
IMatches matches;
Action_newRule newRule = Action_newRule.Instance;
for(int i = 0; i < n - 2; i++)
{
matches = newRule.Match(procEnv, 1);
newRule.Modify(procEnv, matches.First);
}
matches = Action_mountRule.Instance.Match(procEnv, 1);
Action_mountRule.Instance.Modify(procEnv, matches.First);
Action_requestRule requestRule = Action_requestRule.Instance;
for(int i = 0; i < n; i++)
{
matches = requestRule.Match(procEnv, 1);
requestRule.Modify(procEnv, matches.First);
}
/**
* The MutexPimped.grg file has been annotated with [prio] to generate very good static
* searchplans, without the need of dynamically generating searchplans.
*
* Otherwise you would use the following code to get better matcher programs:
*
* graph.AnalyzeGraph();
* LGSPAction[] newActions = actions.GenerateSearchPlans("takeRule", "releaseRule", "giveRule");
* LGSPAction takeRule = newActions[0];
* LGSPAction releaseRule = newActions[1];
* LGSPAction giveRule = newActions[2];
*
* Alternatively you could also write:
*
* graph.AnalyzeGraph();
* actions.GenerateSearchPlans("takeRule", "releaseRule", "giveRule");
* LGSPAction takeRule = actions.GetAction("takeRule");
* LGSPAction releaseRule = actions.GetAction("releaseRule");
* LGSPAction giveRule = actions.GetAction("giveRule");
*/
Action_takeRule takeRule = Action_takeRule.Instance;
Action_releaseRule releaseRule = Action_releaseRule.Instance;
Action_giveRule giveRule = Action_giveRule.Instance;
for(int i = 0; i < n; i++)
{
matches = takeRule.Match(procEnv, 1);
takeRule.Modify(procEnv, matches.First);
matches = releaseRule.Match(procEnv, 1);
releaseRule.Modify(procEnv, matches.First);
matches = giveRule.Match(procEnv, 1);
giveRule.Modify(procEnv, matches.First);
}
int endTime = Environment.TickCount;
PrintResults(endTime - startTime, graph);
// example for reading annotations at API level
Action_annotationTestRule annotationTestRule = Action_annotationTestRule.Instance;
foreach(KeyValuePair<string, string> annotation in annotationTestRule.rulePattern.Annotations)
{
Console.WriteLine("The rule " + annotationTestRule.Name + " is decorated with an annotation " + annotation.Key + " -> " + annotation.Value);
}
foreach(IPatternNode node in annotationTestRule.RulePattern.PatternGraph.Nodes)
{
foreach(KeyValuePair<string, string> annotation in node.Annotations)
{
Console.WriteLine("The pattern node " + node.Name + " of rule " + annotationTestRule.Name + " is decorated with an annotation " + annotation.Key + " -> " + annotation.Value);
}
}
IAnnotationTestNode testNode = graph.CreateNodeAnnotationTestNode();
foreach(KeyValuePair<string, string> annotation in testNode.Type.Annotations)
{
Console.WriteLine("The node type " + testNode.Type.Name + " is decorated with an annotation " + annotation.Key + " -> " + annotation.Value);
}
}
/// <summary>
/// Executes the Mutex benchmark with a problem size of n.
/// This variant uses BaseActions.ApplyGraphRewriteSequence.
/// Although this provides the fastest way to program rule applications
/// for simple scenarios, it needs aprox. 12% more running time than
/// via Apply or Match/Modify.
/// </summary>
/// <param name="n">The problem size, i.e. the number of Process nodes</param>
static void AlternativeThree(int n)
{
int startTime = Environment.TickCount;
MutexGraph graph = new MutexGraph();
MutexPimpedActions actions = new MutexPimpedActions(graph);
LGSPGraphProcessingEnvironment procEnv = new LGSPGraphProcessingEnvironment(graph, actions);
LGSPNode p1 = graph.CreateNodeProcess();
LGSPNode p2 = graph.CreateNodeProcess();
LGSPEdge n1 = graph.CreateEdgenext(p1, p2);
LGSPEdge n2 = graph.CreateEdgenext(p2, p1);
procEnv.ApplyGraphRewriteSequence("newRule[" + (n - 2) + "] && mountRule && requestRule[" + n
+ "] && (takeRule && releaseRule && giveRule)[" + n + "]");
int endTime = Environment.TickCount;
PrintResults(endTime - startTime, graph);
}
/// <summary>
/// Executes the Mutex benchmark with a problem size of n.
/// This variant uses IAction.Apply and IAction.MinMax to apply the rules
/// (also available IAction.ApplyStar and IAction.ApplyPlus).
/// These functions make it very easy to program complex rule applications,
/// while providing the same speed as using IAction.Match and IAction.Modify
/// (this example is not a complex scenario).
/// </summary>
/// <param name="n">The problem size, i.e. the number of Process nodes</param>
static void AlternativeTwo(int n)
{
int startTime = Environment.TickCount;
MutexGraph graph = new MutexGraph();
MutexPimpedActions actions = new MutexPimpedActions(graph);
LGSPGraphProcessingEnvironment procEnv = new LGSPGraphProcessingEnvironment(graph, actions);
LGSPNode p1 = graph.CreateNodeProcess();
LGSPNode p2 = graph.CreateNodeProcess();
LGSPEdge n1 = graph.CreateEdgenext(p1, p2);
LGSPEdge n2 = graph.CreateEdgenext(p2, p1);
Action_newRule.Instance.ApplyMinMax(procEnv, n - 2, n - 2);
Action_mountRule.Instance.Apply(procEnv);
Action_requestRule.Instance.ApplyMinMax(procEnv, n, n);
Action_takeRule takeRule = Action_takeRule.Instance;
Action_releaseRule releaseRule = Action_releaseRule.Instance;
Action_giveRule giveRule = Action_giveRule.Instance;
for(int i = 0; i < n; i++)
{
takeRule.Apply(procEnv);
releaseRule.Apply(procEnv);
giveRule.Apply(procEnv);
}
int endTime = Environment.TickCount;
PrintResults(endTime - startTime, graph);
}
