Re: RNF7535: varchar not compatible with options(*varsize)?

[Dan <dan27649@xxxxxxxxx>]

Thank you, Charles!  Your explanation on *VARSIZE makes sense.  Also, I had
tried previously to use literals in the hash call, but that gave a compiler
error.  CONST solved that issue.

Also, thanks to Jon and Barbara for the explanations they provided!

- Dan

On Mon, Apr 2, 2018 at 9:41 AM, Charles Wilt <charles.wilt@xxxxxxxxx> wrote:

> Dan,
>
> The only parm you can legitimately use options(*VARSIZE) on is InputData
> (if you'd decided to define it as char(16773000)) and ErrorCode
>
> The key thing with  options(*VARSIZE), is that there has to be some way for
> the callee to know how much data was actually passed.
>
> InputData has a matching length parm, LengthOfInputData.  ErrorCode has a
> structure in which the first 4 bytes tell the callee how much space is
> available.
>
> I'd do the PR like so:
>
> dcl-pr hash extproc('Qc3CalculateHash');
>
>  InputData                       pointer value;
>
>  LengthOfInputData               int( 10)   const;
>  InputDataFormatName             char(   8) const;
>  AlgorithmDescription            char(  20) const;
>  AlgorithmDescriptionFormatName  char(   8) const;
>  CryptographicServiceProvider    char(   1) const;
>  CryptographicDeviceName         char(  10) const;
>  Hash                            char(  16) ;
>  ErrorCode                       char( 300) options(*varsize);
> end-pr;
>
> You don't need any of the CONST, but I always use them to indicate which
> parms are INPUT only.  Plus they'd allow you to code the call like so
>
> hash( %addr( DataToHash : *DATA) : inputlen: 'DATA0100':
>       QC3D0500: 'ALGD0500': cryptoserv:
>       cryptoDev: HashValue: errorCD);