Handling exceptions in different units

Hi Dmitriy,



Currently, exceptions cannot be caught if they're thrown in another unit. This is a limitation of the Microsoft JScript engine. We have a suggestion to correct this, and your request has increased its rating.

Thanks a lot for your answer. I have to catch all exceptions in corresponding functions then. Hopefully, we could catch exceptions thrown in other units later. :)

Forgive me for posting twice. But I have the next problem...



I successfully cought exceptions caused by application's crash. Here this the code sample where I try to terminate process (if necessary) and stop test item run.





function IsProcessExist()

{

  var ProcessName = Project.Variables.ProcessName;

  Log.Message("Checking process existance");

  var ProcessExistance = Sys.WaitProcess(ProcessName, 0).Exists;

  Log.Message("Process existance is checked");

  return ProcessExistance;

}



function Terminate(Exception)

{

  var ErrorMessage = new String();

  if (arguments.length == 0)

    var ErrorMessage = "Unknown error or exception";

  else

  {

    if (typeof(Exception) == "object")

      ErrorMessage = Exception.description;

    else

      ErrorMessage = Exception;

  }

    

  var ProcessName = Project.Variables.ProcessName;

  var TerminationTimeOut = Project.Variables.TerminationTimeOut * 1000;     // in milliseconds

 

  Log.Error("Termination reason: " + ErrorMessage);

 

  /* Checking application's process for existance */

  if (!IsProcessExist())

  {

    Log.Message("Application's process is already terminated");

  }

  else

  {

    Log.Message("Attempt to terminate process...");

    TestedApps.ScanMagic.Terminate();       // sending instructions to terminate process

  }

 

  aqUtils.Delay(TerminationTimeOut);

  Log.Message("Stopping test case execution...");

  Runner.Stop(true);          // stopping current test only

}





This one works fine in most cases but sometimes application crashes with "Microsoft Visual C++ Runtime Library Error" window. According to received logs the application's process is already terminated but any actions, for example, aqUtils.Delay(...) or Runner.Stop(true) can't be performed before I activate any other application or press Windows' "Start" button manually. Only after this script continue it's work and stops.



I tried to simulate "Start" button pressing, and received click events in my log...but in real nothing has changed - application's window and modal error window described higher are still preventing normal script execution...



What's the reason? Can I solve this problem by any methods?



Thanks...

Hi Dmitriy,



When your application crashes, it becomes inaccessible for TC because its process gets blocked. If you try to terminate it from script, TC will be unable to do this correctly. As a workaround, you can try closing the crash dialog from an external script. You can find an example here.

Using external script to handle error windows is what I exactly need!

Thank you so much for your invaluable help, Jared!!!

Hi,



I just wanted to second the request to be able to caught exceptions thrown in another unit. We have our test functionality in different units and we need to able to catch exceptions to recover the UI in case of errors. We have long running tests (>100 test items, >10 hours of execution) and every time that there is an error and we can't recover the UI we loose a day of testing.



Thank you,

Patricio.

Hi Patricio,



I've increased the suggestion's rating.

Forgive me for posting here again. But my problem with closing all error messages is still actual. I am using external JS-script proposed higher with little modifications to handle more than one type of window. It works perfect with first type of windows (Visual Studio just-in-time debugger errors) in wndCaptions array but can't handle "Microsoft Visual C++ Runtime Library" windows though I thought it had worked before (I've attached it's screenshot). I determined that the script even couldn't find this window - warning message posts to TC's log only after I press "Ok" button by myself or activate any other application (error window disappears then).

I also read that using JS-script is obsolete and Runner.CallObjectMethodAsync() method could be used. But I still can't imagine how to apply this one to all my tests and all methods that could cause unexpected errors.

At the moment, I have no ideas of JS-script improvement or anything else... Probably, another external application should be written to attempt handling error windows.



P.S. Can you move the topic to another forum? I was inattentive posting this one (didn't see it's only for web testing).





var tcObj, wshShell, wndCaptions;



wndCaptions = new Array();

wndCaptions.push("Visual Studio");

wndCaptions.push("Microsoft Visual C++");



try

{

    tcObj = GetObject("", "testcomplete.testcompleteapplication");

}

catch(e)

{

    tcObj = null;

    //return;

}



if (tcObj != null)

{

    wshShell = new ActiveXObject("WScript.Shell");

    while (tcObj.Integration.IsRunning())

    {

        for (var i = 0; i < wndCaptions.length; i++)

        {

            if (wshShell.AppActivate(wndCaptions))

            {

                /* Registering error window in TC's log */

                tcObj.Integration.GetObjectByName("Log").Warning("\"" + wndCaptions + "\" was found in error window's caption");

                wshShell.AppActivate(wndCaptions);

                WScript.Sleep(1000);

                wshShell.SendKeys("{ENTER}");

            }

        }

    }

}

Hi,



Does it help if you use the full caption of this window (Microsoft Visual C++ Runtime Library)?

Hi, Jared!



No, it doesn't. According to Windows Script Host reference not full title is acceptable if it begins or ends by this string:

http://msdn.microsoft.com/en-us/library/wzcddbek%28VS.85%29.aspx

...and the script really executes the following commands and logs this caption to TC's log but only after human's actions.



By the way, returning to handling exceptions in different units...

My tests are data-driven and, in general, have the following view:





function SCM0025()

{

  Application.Run();   // starts the application using "Application" unit function

  var R;

  var DataDrivenTest = new CDataDrivenTest("testcase0025.csv");

  while (!DataDrivenTest.EOR())

  {

    R = DataDrivenTest.GetRow();

    try

    {

      // iteration commands called from different units

    }

    catch (e)

    {

      Log.Error(e.description);

    }

  }

  DataDrivenTest.Finish();

}





CDataDrivenTest is my own class that uses DDT object to read CSV-files. Due to we can't handle exceptions in different units by general "try...catch" construction I have to put this one to most functions in other units, for example:



// Another unit

function DoSomething()

{

  try

  {

    // we are doing something here

  }

  catch (e)

  {

    Application.Terminate(e);

  }

}



If an exception occurs Terminate() function from "Application" unit will be called. I close/terminate the application there if it's still alive, and then stop current test execution by Runner.Stop(true) call. It is good to continue project's testing with next test item. But I would execute the failed test with next data rows from CSV-file. Probably, we could add some project's variables to store test's last iteration number. But how could we re-run the failed test again saving project's test items order?