Re: [sv-ec] Query related to virtual methods in a class

Don,

Until a pure virtual method is overridden, all class types in
the inheritance chain must be abstract (i.e. be "virtual class").
So derived classes may or may not override a pure virtual
method but if they don't, they must remain abstract.

module top;
    virtual class base;
       pure virtual task t;
       virtual task t2;
           $display("in base::t2");
       endtask
    endclass

    virtual class mid extends base;
    endclass

    class bot extends mid;
       task t;
          super.t;
           $display("in bot::t");
       endtask
    endclass
endmodule

It would be an error if "mid" was not an abstract class
since the method "t" is still pure virtual in mid.  Once
"bot" overrides "t", all methods have implementations and
"bot" is no longer required to be abstact.


If base::t was not pure, then "mid" would not have to
be abstract.  The override in bot would still be valid.

Basically the rule is that not specifying anything in
an intermediate class with respect to a method does not
impact the method inheritance from the base class
(pure virtual remains pure virtual, etc).

Gord.


Don Mills wrote:
> I have second question on the same train of thought that Kakoli 
> presented.  If a virtual base class contains a virtual method, then when 
> the base class is extended, is the extended child class required to 
> include the virtual method?  Or is the virtual method only a prototype 
> if the child class wished to use that method?  I assumed that base class 
> virtual methods were required to exist and be defined in extended 
> classes, but that is not the case with all simulators.  Is my 
> understanding wrong here?
> 
>  
> dm
> 
> 
>     -----Original Message-----
>     From: Kakoli Bhattacharya
>     Sent: Oct 19, 2006 6:07 AM
>     To: sv-ec@server.eda.org
>     Subject: [sv-ec] Query related to virtual methods in a class
> 
> Hello,
> 
> Consider the following code:
> 
> virtual class Parent;
> integer it;
> int i;
> virtual task virtual_task(input reg rg);  //This is the declaration
> endtask
> endclass
> 
> class Child1 extends Parent;
> shortint si;
> longint li;
> virtual task virtual_task1(input reg rg1);
> endtask
> endclass
> 
> class Child2 extends Child1;
> real r1;
> shortreal sr1;
> time t1;
> realtime rt1;
> task virtual_task(input reg rg1);  // This is the redefinition
> it = rg1;
> i = rg1;
> si = rg1;
> li = rg1;
> endtask
> enclass
> 
> My problem concerns the virtual task 'virtual_task'.
> When it has been declared in the parent class 'Parent' then it takes 'rg' as its argument.
> During its redefinition in the child class 'Child2' it takes 'rg1' as its argument.
> Is it necessary that the names of the variables also have to be identical?
> 
> In LRM (1800-2005) Section 7.19 it is stated that "Virtual methods provide prototypes for subroutines,
> i.e., all of the information generally found on the first line of a method declaration: 
> the encapsulation criteria, the type and number of arguments, and the return type if it is needed.
> Later, when subclasses override virtual methods, they must follow the prototype exactly."
> 
> Prototype however do not mean identical variable name.But later on in the same section it is also stated that 
> "Thus, all versions of the virtual method look identical in all subclasses"
> 
> So what does the word 'identical' exactly mean overhere? Does it imply identical argument name also?
> 
> Thanks,
> Kakoli
> 
> 
> 
> ==========================================================
> Don Mills
> LCDM Engineering     (Logic, Coding, & Design Methodology) 
> mills@lcdm-eng.com                        www.lcdm-eng.com
> ==========================================================
> 

-- 
--------------------------------------------------------------------
Gordon Vreugdenhil                                503-685-0808
Model Technology (Mentor Graphics)                gordonv@model.com