Re: EXTERNAL: Re: [vhdl-200x] Update to proposal for arbitrary integers

On 29/10/14 22:43, Trond Danielsen wrote:

> If you open up the language to Nullable types you open up a giant can of
> worms and errors.

I agree.

 > If this sort of functionality in VHDL is need I
> suggest that it is based on the option type known from functional
> programming languages such as Haskell. Only then will you be able to
> rely on the compiler avoid null-exceptions. Of cause a Maybe (Haskell)
> or Option (F#) type would require significant modifications to the VHDL
> type system, but it opens up a whole new world of possibilities.

Currently (and even with vhdl87), you can write:

   type maybe_integer is record
     val : integer;
     is_nan : boolean;
   end record;

And then declare "+", "-", "/" ...  functions to have an integer+Nan
semantic.

If then you add conversion functions like:
   function "+" (l : maybe_integer) return integer is
   begin
      assert l.is_nan = false;
      return l.val;
   end "+";

   function "+" (l : integer) return maybe_integer is
   begin
     return (l, false);
   end "+";

you have the Maybe semantic of Haskell (or BSV).  This is a little
bit heavier to declare in VHDL than in Haskell, but with a generic
package this is doable.

So I'd also prefer not to shake-up the language for the NaN need.

Tristan.


-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.