DSXchange

Hi All,

In many of the date conversions, i have seen the usage of ICONV/OCONV functions as the solutions.

But when i am developing a parallel job, we don't have the option of above Functions in Transformer stage, where as in server jobs we have the same in the transformer as well in the sequence jobs.

Is there any way to implement ICONV/OCONV functions in the parallel jobs in the transformer stage or we need to use a basic transformer to use this function?

Please let me know your views.

You need to use BASIC Transformer stage if you want to use Iconv() or Oconv() functions.

They are simply not available in the parallel Transformer stage, and never will be.

You need to learn the Parallel date handling functions, of which there are many. Primarily look into StringToDate() and DateToString() but a quick check of "Appendix B. Parallel Transform functions" in the Parallel Job Developer's Guide should be high on your list as well.

Hi Ray/Chulett,

Thanks for the information.
I am really learning most of the things from this forum.

Thanks once again for your time.

ray.wurlod wrote:They are simply not available in the parallel Transformer stage, and never will be. ...

Hi Ray, I was hoping to see these functions in parallel transformer one day, just like ereplace made it in 9.1

Is there any particular reason why you mentioned that they will never be available?

The ICONV and OCONV functions converted data to and from a semi-proprietary internal storage format. Though the internal format made some data manipulations easier (like comparing dates), it takes CPU cycles to do the conversions back and forth.

DataStage PX doesn't use that methodology, so it avoids the overhead of the conversions.

To clarify what Andy has written, the engines (server and parallel) use different internal storage mechanisms.

There's nothing to prevent you from creating parallel routines that perform Iconv() or Oconv() functionality, but that would be re-inventing the wheel - in general there are other ways to achieve the same results in parallel jobs.

Thanks for the info, Ray & Andy!