DSXchange

@DATE and @TIME are system variables whose values are assigned when the job starts. As Arnd notes the intrinsic functions Date() and Time() give access to the system clock. Oconv(Date(),"D-YMD[4,2,2]") : " " : Oconv(Time(),"MTS:")

There is one separate Repository per project, at least in version 7 and earlier. Think of the Repository as a schema in a database, for that's really what it is in these versions. In version 8 and later you get a common Repository across all projects (indeed across all products; it is called Informa...

What says what doesn't exist? Preserve the data and control files and run the bulk loader with your DBA looking on. Perhaps there are some trace tools available (to the DBA) that would help. The bulk load is initiated by DataStage, but once it's running DataStage has no more control. You have to dia...

"it was felt"?!! Shades of the old "statistics have shown"! By whom was it felt? Based upon what? Have you designed/conducted any tests of your own? I have no information, but would be surprised if DataStage sort could beat CoSort. It's known that DataStage tsort operator is faster than a UNIX sort,...

Dont try to write a null key to a hashed file.

The code is to be found in the Job Control tab of the job properties, once the job sequence has been compiled. The generated code of a job sequence is read-only, so that the nexus between the generated code and the graphical design from which it was generated is preserved. You could, of course, use ...

In a parallel job, no.

No passive stage is permitted to have an input link and an output link (not counting reject links). This is an absolute rule in parallel jobs.

I fail to see how even a Transformer stage could recognize duplicates "across different runs". Can you please elaborate on that?

Meanwhile, the Sort stage itself (or input link sort) allows you to perform a unique sort. Would that do it for you?

Performance Tip
Putting the Left() function result into a stage variable would mean that you execute it once per row rather than twice per row. For squillions of rows this can make a noticeable difference.

In a word, no. A server job is a loop in its own right (loop until no more data). Any routine can include as many loops as you need, but routines that are executed for every row processed really should be executed as quickly as possible. Perhaps you could post a particular requirement, which we coul...

Using the automatic feature, no.
By designing it yourself or coding it yourself, yes.

If you want to replace the default when running the job, use the -param option. dsjob -run -param Log_Dir=/user/spool/myproject/logdir projname jobname To recover the value that was used, whether it was substituted for $PROJDEF or in the run request, you need to look at the job log. The "job started...

I should have thought #$EnvironmentVariable# is what's required - which would also explain the 'unexpected "$"' message.

Which scripts are you talking about? UNIX shell scripts that ran the job via dsjob, UNIX shell scripts that were run by the job, osh scripts that were created by compiling the job, or some other scripts? Until we know that it will be impossible to provide any meaningful answer. However, the first re...

All fields on a reference input should be nullable if the Lookup Failed rule is Continue. Similarly all fields on the output link that are derived from the reference input should be nullable. Then handle the nulls downstream. If there are any NULL values on the stream input to the Lookup stage one w...