I'm probably overthinking this, but I have a question about getting intellisense to work within my project. I support integration with 3 different products for credit card payment services. Right now, the code is very messy to work through these, and I am refactoring. My goal is to create a single payment interface class that works behind the scenes to create the right object and allow transactions to be processed.

I have created a base class for a transaction. It includes properties for all of the information and empty methods for all the different transaction types.

I have created 3 classes based on the base class which include the actual processing methodologies for each different integration.

I have created a master transaction class that includes a property (initialized to null) for a transaction object. The Init() looks at the system configuration, determines the appropriate integration method, and then creates the appropriate object from the 3 integration classes.

Once the master class is instantiated, I can assign the various transaction properties and call the transaction methods from my application without having to worry about what integration methods are selected.

My question is: Is there a way to do this so that I can have intellisense active on the transaction property. Because it is initialized as null, it has no properties. I could create it doing an ADD OBJECT O AS BASECLASS, but then I cannot reassign it to one of the 3 integration class in my init.

If it helps, I have a set of sample programs illustrating the concept.
testbase.prg defines the base class
testclass1.prg subclasses testbase and represents the first method performing a task
testclass2.prg subclasses testbase and represents the second method for performing a task
testmaster.prg defines a class I can call to perform my task. In the init it determines which method to use and creates an appropriate instance of testclass1 or testclass2.
testcode.prg contains code that I use to execute task.

The object is to be able to type "o.oTestObject." in testcode and have intellisense for that objects properties and methods.

testbase.prg:

DEFINE CLASS TestBase as Custom
  testVar1 = ""

  FUNCTION DisplayTestVar
  ENDFUNC 
ENDDEFINE

testclass1.prg:

DEFINE CLASS TestClass1 as TestBase OF TestBase.prg
  FUNCTION DisplayTestVar
    WAIT WINDOW this.testVar1
  ENDFUNC
ENDDEFINE

testclass2.prg:

DEFINE CLASS TestClass2 as TestBase OF TestBase.prg
  FUNCTION DisplayTestVar
    MESSAGEBOX(this.testVar1)
  ENDFUNC
ENDDEFINE

testmaster.prg:

DEFINE CLASS TestMaster as Custom

 #if (.f.)
   LOCAL this as TestMaster OF TestMaster.prg
 #endif

   oTestObject = null
 
   PROTECTED nMethod 
   nMethod = 0
   
   FUNCTION Init
   
*
* figure out which method to use - for the sake of demonstration I will hard code a selection
*     
    this.nMethod = 2
    
    DO CASE 
      CASE (this.nMethod = 1)
        this.oTestObject = NEWOBJECT("testclass1", "testclass1.prg")
        
      CASE (this.nMethod = 2)
        this.oTestObject = NEWOBJECT("testclass2", "testclass2.prg")
        
    ENDCASE
    
  ENDDEFINE

testcode.prg:

LOCAL o as TestMaster OF TestMaster.prg

o = NEWOBJECT("testmaster", "testmaster.prg")
o.oTestObject.testVar1 = "Hello World"
o.oTestObject.DisplayTestVar()