floatfixlib synthesis

I ve written a code below which can be simulated but cannot be synthesized in Leonardo Spectrum ver. 2007a.37

Here is the code

library IEEE;
use IEEE.std_logic_1164.all;
use IEEE.numeric_std.all;
library floatfixlib;
use floatfixlib.math_utility_pkg.all;
use floatfixlib.fixed_pkg.all;

entity test is
	port(a,b: in sfixed(7 downto -6);
	     c: out sfixed(8 downto -6));
end test;

architecture testing of test is
begin

    c <= a+b;      

end testing;

Here is the synthesis error

-- Reading vhdl file C:/Documents and Settings/batur/Desktop/test.vhd into library work
"C:/Documents and Settings/batur/Desktop/test.vhd",line 5: Warning, math_utility_pkg is not declared in library floatfixlib.
"C:/Documents and Settings/batur/Desktop/test.vhd",line 6: Warning, fixed_pkg is not declared in library floatfixlib.
"C:/Documents and Settings/batur/Desktop/test.vhd",line 9: Error, sfixed is not a known type.
"C:/Documents and Settings/batur/Desktop/test.vhd",line 10: Error, sfixed is not a known type.
"C:/Documents and Settings/batur/Desktop/test.vhd",line 9: Error, sfixed requires 0 index values.
"C:/Documents and Settings/batur/Desktop/test.vhd",line 10: Error, sfixed requires 0 index values.
Error in file C:/Documents and Settings/batur/Desktop/test.vhd.
-- Error found in VHDL source

It says library error. Is Fixed point library "floatfixlib" synthesizable or not???
How can we write fixed point code that is synthesizable?

universeee