Doubt on Scope Randomize Function

SystemVerilog
,
1 

LRM  18.12  says :: 

The scope randomize function, std::randomize() , enables users to randomize data in the current
scope without the need to define a class or instantiate a class object

But the LRM isn’t clear if an object with random properties can be used as argument to it

Example ::


class seq ;

     rand bit [1:0] b ;
          
	  bit [1:0] a ;

  endclass

  seq  obj ;

  initial begin

      obj = new();

    if ( std::randomize(obj.a) with { obj.a == 3 ; } ) //  Valid ??   
      begin
         $display("Success with a == %0d",obj.a);
      end
    else
     begin
        $display("Fails");
     end

    if ( std::randomize(obj.b) with { obj.b == 2 ; } ) //  Valid ??
      begin
         $display("Success with b == %0d",obj.b);
      end
    else
     begin
        $display("Fails");
     end
  end

I see warning message on some Simulators ::

Warning: (vopt-2961) Argument #1 for std::randomize() function is non-LRM compliant.

While some throw Compilation Error ::

**if ( std::randomize(obj.a) with { obj.a == 3 ; } )

xmvlog: *E,RNDRARG : The argument to class or scope randomize must be a simple identifier of integral type.
**

[Q] Is this a violation based on LRM ??

2

In reply to TC_2017:

This is an open issue in the LRM. 0002967: define explicitly legal set of data types for randomize variables list - Accellera Mantis

Hopefully the IEEE P1800 committee can address this in the next revision of the standard.

3

In reply to dave_59:

It would also be nice to randomize an element of a struct without having the compiler yell at you.


struct
{
  int a;
  int b;
} some_struct;
  
initial begin
   void'(std::randomize (some_struct.a));
end

produces

** Warning: testbench.sv(12): (vlog-2961) Argument #1 for std::randomize() function is non-LRM compliant.

To work around you need to introduce a dummy variable.


initial begin
   int dummy_var;
    
   void'(std::randomize (dummy_var));
   some_struct.a = dummy_var;
end

So much typing!

4

In reply to sbellock:

https://accellera.mantishub.io/view.php?id=2970