RE: [vhdl-200x-ft] Generic packages and Generic Subprograms

Jim and Karl,

Funny how we sometimes go around in circles! My original proposal had the
generic clause in subprograms following the subprogram name. However, since
a parameter is different from a port, I proposed the optional keyword
"parameter" to introduce the parameter list. The syntax was

  subprogram specification ::=
      procedure designator
          [ generic ( generic_list ) ]
          [ [ parameter ] ( formal_parameter_list ) ]
   | [ pure | impure ] function designator
          [ generic ( generic_list ) ]
          [ [ parameter ] [ ( formal_parameter_list ) ] return type_mark

However, committee members decided they didn't like the syntax, and
preferred to have the generic clause precedure the rest of the subprogram
specification. I think Jim has raised a valid point about the "Hamming
distance" between different constructs, so the committee may wish to
reconsider the syntax. I personally prefer my original proposal, on the
grounds of consistency among constructs that have a generic list and a
port/parameter list. Jim's comments strengthen my preference.

Cheers,

PA

--
Dr. Peter J. Ashenden                        peter@ashenden.com.au
Ashenden Designs Pty. Ltd.                   www.ashenden.com.au
PO Box 640                                   Ph:  +61 8 8339 7532
Stirling, SA 5152                            Fax: +61 8 8339 2616
Australia                                    Mobile: +61 414 70 9106
> -----Original Message-----
> From: owner-vhdl-200x-ft@eda.org 
> [mailto:owner-vhdl-200x-ft@eda.org] On Behalf Of Karl Eisenhofer
> Sent: Friday, 24 December 2004 06:07
> To: Jim Lewis
> Cc: vhdl-200x-ft@eda.org
> Subject: Re: [vhdl-200x-ft] Generic packages and Generic Subprograms
> 
> 
> To be more consistent with the entity generic declarations, it seems 
> that the generic declaration should not appear before the subprogram 
> declaration.  However, the generic names need to be 
> introduced prior to 
> the parameter list.  Therefore, I suggest putting the 
> subprogram generic 
> declarations just after the function name, but just before 
> the parameter 
> list as so:
> 
> function a_generic_fun generic (type fun_type) (L: fun_type) returns 
> fun_type;
> 
>  From a new user point of view, the syntax should ideally be 
> consistent 
> for all generic/port declarations, i.e.
> 
> entity <name>
>     [generic <list>]
>     [port <list>]
> is
> ...
> 
> package <name>
>     [generic <list>]
> is
> ...
> 
> function <name>
>     [generic <list>]
>     [[port] <list>]            -- optional port keyword, 
> allows current 
> function syntax
>     returns <type>
> is
> ...
> 
> procedure <name>
>     [generic <list>]
>     [[port] <list>]
> is
> ...
> At first glance it would seem to be possible to support both current 
> syntax (wherein for entities, the generic and port lists come 
> after the 
> "is") and this proposed syntax for backward compatibility. 
> 
> Thoughts?
> 
> Karl
> 
> Jim Lewis wrote:
> 
> > Peter,
> > With the generic clause following the generic package and
> > the generic clause preceding the generic 
> procedure/function, seems to 
> > lead to some interesting syntax errors.
> >
> >
> > Case 1:
> > Generics on package and subprogram (no issues, baseline case)
> > ------------------
> >
> > package a_generic_pkg is
> >   generic(type pkg_type) ;
> >
> >   generic(type fun_type)
> >   function a_generic_fun(L: pkg_type) return fun_type ;
> > end a_generic_pkg ;
> >
> >
> > Case 2:
> > Generics on package and but not on subprogram
> > Issue:  If leave off ";", then the generic belongs to
> >         the function and not the package.
> > ------------------
> > library ieee ;
> > use std_logic_1164.all ;
> > package a_generic_pkg is
> >   generic(type pkg_type) ;
> >
> >   function a_generic_fun(L: pkg_type) return std_logic_vector ; end 
> > a_generic_pkg ;
> >
> >
> > Case 3:
> > Generics not on package and on subprogram
> > Issue:  If add ";", then the generic belongs to
> >         the package and not the function.
> > ------------------
> > library ieee ;
> > use std_logic_1164.all ;
> > package a_normal_pkg is
> >
> >   generic(type fun_type)
> >   function a_generic_fun(L: fun_type) return std_logic_vector ; end 
> > a_normal_pkg ;
> >
> >
> >
> > At one point in time I think there was a goal to keep
> > a hamming distance between correct functionality.
> > Should we consider moving the generic clause for the
> > package before the package like the subprogram or
> > alternately should we consider moving the generic
> > clause to after the subprogram declaration.
> >
> > Cheers,
> > Jim
> 
>