Re: [vhdl-200x] VHDL-2008: Records and aggregates

Hi Tristan,
Here are a few more details, the following are defined in a package.    For the whole package, see: http://www.synthworks.com/downloads/AggregateIssuePkg.vhd

   type UartStimType is record
     Data         : std_logic_vector(7 downto 0) ;
     ErrorMode    : std_logic_vector(3 downto 1) ;
   end record ;

   function to_string (
     constant  rec       : in    UartStimType
   ) return string is
   begin
     return "Data = " & to_hstring(rec.Data) &
            ", Parity Error: " & to_string( rec.ErrorMode(1)) &
            ", Stop Error: " & to_string( rec.ErrorMode(2)) &
            ", Break Error: " & to_string( rec.ErrorMode(3)) ;
   end function to_string ;

   function to_string (
     constant  Data      : in    std_logic_vector(7 downto 0) ;
     constant  ErrorMode : in    std_logic_vector(3 downto 1)
   ) return string is
   begin
     return to_string( (Data, ErrorMode) ) ;
   end function to_string ;

In the second function, the return statement:
     return to_string( (Data, ErrorMode) ) ;

With VHDL-2008 is now ambiguous since "(Data, ErrorMode)" can be interpreted as a record, of a type such as UartStimType, or it can be interpreted as a single std_logic_vector.

Best Regards,
Jim

On 4/10/2015 11:57 AM, Tristan Gingold wrote:
> On 10/04/15 18:32, Jim Lewis wrote:
>> Hi,
>> I just ran into an interesting issue with VHDL-2008 aggregates. In
>> VHDL-2008, we allowed aggregates to effectively do concatenation when
>> the target is an array.
>>
>> Unfortunately this means you have a to_string function that is
>> overloaded to handle records, it is now ambiguous since the record
>> aggregate can also be interpreted as std_logic_vector.
>>
>>        write (buf,  to_string( (UartRxRec.Data, UartRxRec.ErrorMode) )) ;
>
> No, this is ambiguous since VHDL 87, as the type of an aggregate is
> determined by the context.
>
>> If you use named aggregates, this is not an impact.  However, if you do
>> not, it may be a significant impact.
>
> No, that's not correct.  Use of association by name has no impact for
> determining the type.
>
> Tristan.
>
>> Hence, now you have to add a type qualifier.  This is an unfortunate
>> side-effect.
>>        write (buf,  to_string( UartStimType'(UartRxRec.Data,
>> UartRxRec.ErrorMode) )) ;
>>
>> The original change was targeted at handling situations like:
>>    (CarryOut, Sum_unsigned) <= ('0' & A_unsigned) + B_unsigned ;
>>
>> And also handles situations like:
>>    (ExtraBit, Accumulate_ufixed) <= Accumulate_ufixed + A_ufixed ;
>>
>> Really wish we had allowed concatenation on the LHS instead.
>>
>> We may need to look into how big of an impact this causes.
>>
>> Jim
>>
>
>

-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Jim Lewis                                  Jim@SynthWorks.com
VHDL Training Expert                       http://www.SynthWorks.com
IEEE VHDL Working Group Chair
OSVVM, Chief Architect and Cofounder
1-503-590-4787
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~


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