DSXchange

If they're tables, as you say, why not simply join them using SQL? If they're not, load two into hashed files, and do standard lookups in a Transformer stage.

There are various techniques, most work best in sequential mode. You can use system variables like @INROWNUM, which always begins at 1. You can use stage variables in a Transformer stage. You can use a Column Generator stage. And so on. There is no technique for remembering anything from yesterday i...

Sparse lookup is only available for DB2 and Oracle.

You need either to pre-load a Lookup File Set with the data in your lookup table, or involve a virtual Data Set. In the latter case you really need to have enough memory available such that the entire reference data set can be resident.

There is a command line interface called dsjob that you invoke from your Perl script.

Read about dsjob by searching the forum, or by reading either the Server Job Developer's Guide or the Parallel Job Developer's Guide. It's also mentioned in the DataStage Operator's Guide.

Oh my God, they've killed Kenny!

Ken or Kenneth, please - he doesn't like the other name.

It's not a tab, it's a drop down list in version 7.0 (which I am using this week). It's enabled on Hashed File stages that are serving reference inputs and it's enabled on Hashed File stages that are serving stream links. NLS is not enabled. As far as I am aware and can discover this is default beha...

I'm doing a lot of those right now, using DataStage 7.0. My preferred technique is to use a stage variable for each pair of column comparisons. This allows me to Trim, pad, do whatever I need to do (for example "nulls compare equal") before determining whether the columns match or not. These Boolean...

By "client" do you mean "customer" or "PC"? If the former, then email (a Notification activity in the controlling job sequence) is the ideal mechanism. If the latter, then notification is right there in Director.

The status (hashed) file is not the same as the log (hashed) file. Clearing the status file will have no impact on a corrupted log file, which may have become corrupted by becoming too large, or by some other mechanism. Clearing the log (hashed) file may help. CLEAR.FILE RT_LOG16 is the command, whi...

Thank you for your kind words. My posts/day rate has fallen over the last year or so, for the very good reason that more others are now able to provide sufficiently good answers to questions, so that I don't need to as often. It used to be over 10 per day. Of course, you'll still get a "full Wurlod"...

Welcome aboard. :D The Lookup stage has specific requirements, in that you need to have set up the Key property for the lookup to occur (and, possibly, whether to perform the lookup at all). Can you draw an ASCII picture of your job design so we have a clearer idea of what you are trying to achieve?...

1. Install DataStage on the UNIX server. 2. Configure all database access on the UNIX server. 3. Make requisite changes to dsenv and .odbc.ini on the UNIX server. 4. Create all required projects on the UNIX server (via DataStage Administrator). 5. Export project from Windows. 6. Import project on UN...

You need to purchase an additional licence to be able to develop parallel jobs.

To be able to compile and run parallel jobs you need an appropriate version. Currently the only version that supports the parallel engine on Windows is 7.5x2.

The fourth argument to DSExecute is the system return code; for SQL it returns the number of records processed or, if a negative integer, an error code. Job control routines of the kind we're talking about are not usually created in the Routines branch but, rather, as code on the Job Control tab of ...

Compile All is still out there, on sites that aren't up to a version of DS where MCM is supported. It behoves us, therefore, to maintain the forum even though there's little traffic on it. The problem with changing categories is who does the labor to move existing posts (there are many of them now) ...