Re: [vhdl-200x-ft] numeric_bit package

Jim Lewis wrote:

> All,
> For types signed and unsigned in numeric_bit, we want the same
> implementation we agreed upon for types signed and unsigned in
> numeric_std.

Agreed.

> Given that they are implicitly defined, we can still overload
> them in numeric_bit. Worst case is we use the rules of FT01, but
> perhaps this one does not require the new rules
> since the types signed and unsigned will be both
> created and overloaded in the same package, numeric_bit.

Yes, but if they are implicit, why not use the ones in the that are
implied?  I have run a testcase to prove the functionality of the
implicit shift operators, and they are working great.

My plans are the following:

1) numeric_bit - remove all shift operators EXCEPT "sra" and "sla",
which I will implement there directly.  NOTE:  These were there
in the original 1076.3 version of "numeric_bit", I am commenting them out.

2) std_logic_1164 - implement "sll", "srl", "ror", and "rol".
Specifically do NOT implement "sra" and "sla" in this package.

3) numeric_bit_unsigned - implement "sra" and "sla" in this package
(treats bit_vector as an unsigned).  Overrides the default functionality.

4) numeric_std - Implement "sla" and "sra" to be mathmetically correct.
"sll", "srl", "ror", and "rol" are already there.

5) numeric_unsigned - implement "sra" and "sla" in this package
for "std_logic_vector".

6) fixed_pkg - same as "numeric_std".

7) fphdl_base_pkg - NO shift operations implemented.


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