DSXchange

Had you mentioned at the beginning that RoutineA was a before/after subroutine and not a transform function a lot of what we went through could have been avoided.

Better than what? What really is your question?

Hashed files are brilliant for key equality lookups because a key can be found with a single logical I/O irrespective of the number of keys in the table. No table scan, no index scan, no index even.

I don't suppose you bothered to Search?

You can test the mapping with an Oconv() function. Create a routine that applies the Iconv() function to its argument with "NLSmapname" as the second parameter to Oconv() then test whether the conversion was successful using the Status() function. Obviously you need to replace "mapname" in the above...

So not all that relevant to an ETL tool then?

Rubbish. You can read the log whenever you wish. So can DS API functions.

Which particular "it"? There have been so many in this thread! Take a step back, and write out a specification in English (with no mention of any ETL tool) what it is you want to accomplish. And let's begin again from there.

Remember that there are NO DATA TYPES in server jobs.

No idea. There's nothing in DataStage per se that would destroy a job log table. Could be vandalism, could be an operating system flaw with file units getting confused, just to mention a couple off the top of my head. The code that drops a job log table could be a malicious routine; the only part of...

There's no such thing. By the way, the correct term is "hashed" file. A hashed file is a database table. Asking for a sample of a hashed file is like asking for a sample of an Oracle or DB2 table. It's just not going to happen. There are both dynamic and static hashed files in each project directory...

Yes, but don't.

If you're using ODBC you must follow the ODBC rules (YYYY-MM-DD for dates), not the Oracle rules (whatever those have been set to).

It's an aid to productivity - you can eliminate low frequency occurrences because you're typically more interested in the higher frequencies, for example to decide whether to apply a rule override. As to WHY you're defining anything, that would require mind reading and I'm not feeling up to doing th...

Did you happen to Search for an answer before posting? I'm sure this has been answered in the past.

There will be two entries in the job sequence log indicating that the routine has been called and that it has returned. Each of these bears a timestamp. The rest is subtraction.

See what you get for not describing the entire scenario? Break your data into two streams based on the value of FLAG. Run each stream through an Aggregator to perform the counts, and bring the streams back together in a Funnel stage. Downstream of that allocate T0FLAG or T1FLAG (in a transformer sta...

Use Compare() function specifying "L" (left-justified) as the third argument.