sign bit for digits type

[luo@cybercom.net ()]

Subject: sign bit for digits type
Newsgroups: comp.lang.ada
Organization: Cyber Access Internet Communications, Inc.
Summary: 
Keywords: 

I have a float type as follows:

package A is
   type float_32 is digits 6;
end A;

This type is unsigned as defined by the language or the compiler?
I declare an object and a '-' operator as follows to handle negative
numbers:

with A;
package B is
  function "-" (x, y : A.float_32) renames A."-" return A.float_32;
  Real :A.float_32 := -5.0;
end B;

What happens in procedure C when I do the following:
with A;
with B;
procedure C is
  value : A.float_32 := B.Real;
begin
  null;
end C;

What is value equal to, 5.0 or -5.0?
Is this compiler dependent?

TIA.