[vhdl-200x-ft] Package layering

Looking at all of the conversion functions, and how they set
on top of one another (use a fixed with font to read this):
Package Dependances
std.standard none
std.vhpi std.standard
std.textio std.standard
ieee.std_logic_1164 std.standard std.textio
ieee.numeric_bit std.standard std.textio
ieee.numeric_bit_unsigned std.standard std.textio
ieee.numeric_std std_logic_1164
ieee.numeric_unsigned std_logic_1164
ieee.fixed_pkg numeric_std
ieee.fphdl_base_pkg fixed_pkg
ieee.fphdl32_pkg fphdl_base_pkg
ieee.fphdl64_pkg fphdl_base_pkg
ieee.fphdl16_pkg fphld_base_pkg
ieee.fphdl_pkg fphdl_base_pkg

The idea being that any fixed point math package will need to also
have integer math in it. Also that any floating point algorithm
will also have fixed point math in it. This allows me to layer
the conversion functions correctly, it is also the way designers
will need to use it.

In the latest revision I am also removing the complex math routines,
as we plan to do something "more elegant" with them at a later time.

Question: I have in the floating point packages several "algorithemic"
funcitons, "sqrt, tan, exp, X**Y", etc. These functions use the
basic functions in some sort of mathematical series to create a
result. The answers are inexact, as I have not mathematically calculated
how many interations on the series I need to use to get closure.
Should I keep the in or loose them? We can always add them back in
in the next pass.

-- 
David W. Bishop dbishop@vhdl.org       All standard disclaimers apply.