[sv-bc] Dotted names within inlined constraints

Sending this to both EC and BC since there is a bit
of cross-over name resolution interpretation here.

I'd like to clarify a couple of details regarding dotted name
resolution form inlined constraints. The key sentence
is in 1800-2009, 18.7:

      Then, if a name fails to resolve within the randomize() with
      object class, the name is resolved normally starting in the
      scope containing the inline constraint.

The given text mostly assumes simple identifiers. What are the
implications for dotted names?

Example 1:

Assume we have the following:

   module child;
     int x;
   endmodule

   module tb1;
     child m();

     class A;
        rand int y;
     endclass
     A a = new;
   endmodule

   module tb2;
     struct { int x; } m;

     initial begin
         m.x = 2;
         tb1.a.randomize() with { y == m.x; };
         $display("%d", tb1.a.y);
     end
   endmodule

My assumption is that "m.x" does not "escape"
out of class A and into a hierarchical resolution from
the point of declaration of class A. This means
that "m.x" refers to tb2.m.x and that the output
of the example is 2.

Example 2:

Now assume we have the following:

   module tb;
     class A;
        rand int y;
        struct { int zed; } m;
     endclass
     A a = new;
   endmodule

   module tb2;
     struct { int x; } m;

     initial begin
         m.x = 2;
         tb.a.randomize() with { y == m.x; };
         $display("%d", tb.a.y);
     end
   endmodule

My assumption is that this is an error since "m" as
a name *is* found in the target object but "x" is not
found as a selected member within that structure.
In this case I follow the precedent of 23.7 and interpret
"the name" as being "the first name in the dotted name".
So m.x does not resolve to be tb2.m.x; the design as given
should produce an error at compile/elab time.

Gord.

-- 
--------------------------------------------------------------------
Gordon Vreugdenhil                                503-685-0808
Model Technology (Mentor Graphics)                gordonv@model.com
-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.