Re: parameter definition in user-defined function

Just had another look at the functions, but there is indeed a restriction 
on using only scalars in user-defined functions as expressed by the LRM 
section 4.6 and the BNF in Annex A of the LRM.

Even though the input/inout/output declarations do not disallow the use of 
array arguments, the declarations for the direction and the type have 
quite different structures:

        `include "disciplines.vams"

        module vsourcea(a, b);
        inout a;
        inout[0:3] b;
        electrical a;
        electrical[0:3] b;

          genvar k;

          analog function real sumarray;
          input[0:3] nums;
          real nums[0:3];

            sumarray = nums[0] + nums[1] + nums[2] + nums[3];

          endfunction

          real nums[0:3];

          analog begin
            for (k = 0; k <= 3; k++) begin
              nums[k] = V(a, b[k]);
              V(a,b[k]) <+ sumarray(nums);
            end
          end

        endmodule

Now compare the heading of the function with that of the module - seems a 
bit odd, I think.

Is there - by the way - any simulator currently around that can correctly 
handle the above (legal!) Verilog-A code?


Marq Kole
Competence Leader Analog Simulation, Philips ED&T


owner-verilog-ams@eda.org wrote on 15-08-2005 17:27:19:
> 
> Having recently tried to get some more complicated code to work in 
> Verilog-A, I had a long and hard look at the user-defined functions.
> There are a couple of issues here that I think need to be addressed: 
> 
<SNIP>
> 
> 3. I have always understood that functions can only accept scalar 
> values, however there is no indication in the BNF that this 
> restriction is real. In calling a function there is no limitation in
> providing an array variable or array parameter.  Moreover, these are
> not limited to scalar entities, but can be arrays as well. Also this
> is not mentioned anywhere in the LRM section 4.6.1. 
> 
> 4. Given that there is no limitation on the arguments to use arrays,
> we should also lift the limitation that a function can only return a
> scalar value, as there is no need to do so. 
>
<SNIP>