Forum Discussion

rmnrdi's avatar
rmnrdi
Contributor
7 years ago
Solved

Handling errors in DDT

I've created a DDT "framework" that I use to pass data into a test template. This is the function that controls the whole thing: // Start Here function Main() { aqPerformance.Start(); //Used...
AlexKaras's avatar
AlexKaras
Icon for Champion Level 2 rankChampion Level 2
7 years ago

Hi,

 

Not sure if I completely got your question... Also I am not quite sure what state you are talking about here: "if the test fails, the test loses it's state".

However, maybe this will help:

-- Your test project must be set up to not stop on error;

-- Each PopulateXXX() test function must return boolean as an indicator of whether or not the execution of the whole test can be continued;

-- Transform your main code to something like that:

  var bResult;
    while (!JEDriver.EOF())
    {  
      Log.Message("******************** Test run number " + i + " has Started... ********************************")
    
      bResult = iv.LoadUIVariables(i);
      if (bResult)
        bResult = SetUIVariables(); 
      if (bResult)
        bResult = PopulateIDPage();
      ...
          
      Log.Message("^^^^^^^^^^^^^^^^^^^^^^^^^^^ Test run number " + i + " has finished. ^^^^^^^^^^^^^^^^^^^^^^^^^^^^");  

      // you may consider whether or not to continue with the next data loop
      if (!bResult)
        break;

      //Indexes to next line of data sources
      i++;
      //Indexes to next line in id file
     JEDriver.Next();       
    }
...

Does this help?

 

 

tristaanogre's avatar
tristaanogre
Esteemed Contributor
7 years ago

What AlexKaras said.  Generally, you've constructed a similar framework to what I've used in the past (Great minds....).  What I've done in addition to what has already been said is the judicial use of try/catch/finally groupings around critical code and frequently using "throw" to exit out of a test case if a critical fail point is reached.  This allows me to "bubble" up a status to the main driver routine as a "false" result so that, if a test case returns "false", we log it as a problem and move on to the next.