Interleaving of instance-based scenarios

Question

Hi,

I would like to create a  machine with two instances of my adapter class and call its methods f() and g() in sequence for each instance. In addition, I would like to have interleaving of f() and g() with respective object instances

I tried something like this:

new MyAdapter();_.f();_.g() ||| new MyAdapter();_.f();_.g()

but the output is not what I am after.

How would I let SpecExplorer generate something like this:

adapter0.f();

adapter1.f();

adapter0.g();

adapter1.g();

Answer 1

Hi Dharma11,

You are not composing the sceanrio from left with a model. This is the usual usage.
Assuming that you just left this away here is a machine you might be looking for:

machine TwoThreadScenario2() : Main
{
    new ComplexThread; new ComplexThread; ((_.f;_.g))
}

machine SlicedThreadModelProgram() : Main
{
    TwoThreadScenario2 || ThreadModelProgram
}

Hope it helps

Answer 2

Hi bububa,

Thanks for your reply. I should have put the model program in my example. I also tried your example. But I have one condition that g() should be called only after f() is called for each instance. What happens is that this machine creates two ComplexThread instances and calls f() on one instance and g() on another instance. I want to avoid this interleaving. Is it possible in Spec Explorer?

Answer 3

Hi Dharma11,

ok, if you need more control from the scenario, you can use the let-statement like this:

http://social.msdn.microsoft.com/forums/en-us/specexplorer/thread/AAA9B38E-6398-47A7-8BFE-C2FBA09A476F

Hope this can solve your problem

Answer 4

Fantastic. It worked with the "let" statement. Great!

Answer 5

Hi,

I have an instance-based model and trying to generate tests with two instances of the class under test. For example, I place my machines below. I am not sure whether there are other (good) ways to use Spec Explorer for testing with multiple instances.

The reason I am asking is that sometimes I want to interleave all actions with two (or more) instances. My machine below used the "|||" operator for this purpose. But the only difference in the operands is the instance id. Just curious how this situation is typically handled in Spec Explorer.

// Test Scenario: Connect and Disconnect on Two instances

machine TwoInstancesConnectDisconnectScenario() : Main where ForExploration = false
{
   let SoftwareBusAdapterImpl conn, SoftwareBusAdapterImpl conn2 in
   
   (new conn.SoftwareBusAdapterImpl;conn.Create;(conn.Connect & conn.Disconnect);conn.Destroy)

   |||

   (new conn2.SoftwareBusAdapterImpl;conn2.Create;(conn2.Connect & conn2.Disconnect);conn2.Destroy)
}

machine TwoInstancesConnectDisconnectProgram() : Main where ForExploration = true
{
    (TwoInstancesConnectDisconnectScenario;exit) || ModelProgram
}

machine TwoInstancesConnectDisconnectTestSuite() : Main where TestEnabled = true
{
    construct test cases where strategy = "shorttests" for TwoInstancesConnectDisconnectProgram()
}

Answer 6

Hi Dharama11,

If you model open systems (dynamic creation/annihilation of state objects),
instance based modeling is perfect due to it's elegant control via cord.
Using scenarios/embedded code you can fully control your exploration reading/writing states, reading/writing in/out parameters, control temporal properties etc.

If you like, you can rewrite your code like this:

machine OneInstancesConnectDisconnectScenario(SoftwareBusAdapterImpl conn) : Main
{ 
   (new conn.SoftwareBusAdapterImpl;
    conn.Create;
    (conn.Connect & conn.Disconnect);
    conn.Destroy)
}

machine TwoInstancesConnectDisconnectScenario() : Main
{
   let SoftwareBusAdapterImpl conn,
       SoftwareBusAdapterImpl conn2 in

   OneInstancesConnectDisconnectScenario(conn) |||
   OneInstancesConnectDisconnectScenario(conn2)
}

Hope this helps.

Answer 7

Hi,

Thanks. Suppose we want to scale-up the above machine for several connections (e.g, 100), is there a way to compactly specify in Cord? By compact I mean not explicitly enumerating machines and scenarios for 1, 2, 3, ... connections. This feature will help to validate performance and timing problems with the SUT.

Regards,

Dharma