DSXchange

Same command.
And a friendly reminder, "No Hijacking of threads."

That should be just fine. Make sure you create a dummy variable and set the constraint to @INROWNUM = 1, so that it does'nt run forever.
The array size wouldnt matter in this case.

Ande, I meant code not the compiler. I am well aware of the fact that a C++ compiler is used.

See if thishelps.

You can write it to NULL. In unix it would be /dev/null. I think its windows equivalent is NUL or \NULL.

Sure, just pass the new value to the stage variable and it will be updated with the new value coming in with every row.

What is different between the two systems? Is this a 64 bit hashed file? Also, what kind of transformations are you doing with the two columns? It will depend if its straight load, or a routine call with complex manipulations. If its none of that, then even 2300 rows per second on your dev box seems...

One of the projects at my current site is implementing something similar. The flow is such that a job keeps doing a select from the trigger table and populating a file. In Job control the contents of the file are checked, if we get what we want, we trigger the jobs, else the job goes to sleep for x ...

Have you tried searching, the manuals, pdf help files

UnixOutput, is this the Output from DSExecute() ?
If yes then it is a dynamic array where every element is seperated by @FM which in reality are the line feeds. If you just want the first element you can just request for the same by using

kumar_s wrote:Oh.. ok, the discussion is for the next method. The usage of filter is coupled with Sort stage. And hence there is no need to use of stage variable as well.

I guess I couldnt gather much at a glimpse.

kumar_s wrote:Any specific reason, if Transformer can be replace by Filter, it should be good isn't?

No reason to add another transformer. In order to use stage variables, usage of transformer is inevitable. Just constraint the output in the same transformer. Hence eliminating the need of a filter stage.

Pay attention, next month Ill go by some other name

Delete previous weeks file in the before job subroutine. As for the naming convention, Ray has given you a solution.