DSXchange

Welcome aboard! :D Given that it's a single column in a sequential file, preprocess using sort with the -k option to sort by that column. It's then very easy to use a pair of stage variables to detect when the value in that column changes. This technique has been presented before on this forum - do ...

CODE=0 is more usually reported as DBMS Code = 0; it's the error code returned from the database server. However in this error message it appears (because there are no square-bracketed tokens beyond [ODBC] that the connection either failed or was rejected by the ODBC driver manager. It appears, inde...

I think the message is quite clear.

Either the directory containing your library is not in the search path (LD_LIBRARY_PATH, LIBPATH or SHLIB_PATH, depending on your UNIX), or the routine was not compiled using a C compiler compatible with that platform (or not the right kind of C compiler).

From memory they did it differently than that. The derivation expression on an output column from a Transformer stage (the column that feeds the fact table) invoked a transform function (Routine) that invoked the SP using BCI functions, and the function returned the generated GUID (cast as VARCHAR(3...

If you can get an ODBC driver for Excel (I know they exist in the Windows environment but am not sure about UNIX), then you create a DSN that refers to the workbook (the ".xls" file). Each worksheet in the workbook is a table. When importing the table definitions, make sure you check the "system tab...

Try searching the forum for VERTICAL PIVOT, as this is what you are trying to accomplish.
It's not a job for the Pivot stage, which performs horizontal pivots.

There's not enough information here for diagnosis. For example, what is the SELECT query for each stage? What colour is the link that isn't green? (By default blue = waiting, red = aborted). If it's waiting, is there a locking issue in Oracle - are you, perhaps, trying to update the same table, or h...

Or you can even use the "rightmost substring" to return the rightmost eight characters (remember that there are no data types within DataStage). There are two syntaxes:

An alternative approach is to use the Fmt() function with a mask. for example

Nothing that's supplied. You could write your own wrapper to dsjob. For example: #!/bin/sh project=myproject # name of your project for jobname in `dsjob -ljobs $project` do jstatus=`dsjob -jobinfo $project $jobname| grep "Job Status" | cut -d: -f2` echo $jobname: $jstatus done

I've seen this done, it was SQL Server 7 (but that shouldn't matter). Don't have the exact details to hand, but I recall that a stored procedure that performed the insert then returned the GUID value was employed. Because this was invoked synchronously (from a Routine iirc) the job ran a tad slower,...

I think there's a special variant for North America called SNAFU.

Search the Forum to determine how to use stage variables to detect whether an input value has changed.
Add to this technique's an expression that checks for 950 and resets a "counter" stage variable. An occurrence of a header line in the input stream will also reset the counter.

It's not in the DataStage documentation how to do this, because the philosophy of DataStage is that projects are completely independent of each other. It is, nonetheless, possible. You need one SDKSequences and one D_SDKSequences file, in any location that the multiple projects can "see". You can us...

That said, the answer is almost certainly the correct one. About the only other possibility is that the shared library search path environment variable (LIBPATH on AIX, LD_LIBRARY_PATH or SHLIB_PATH on other operating systems) is not correctly set up. Please don't request my services explicitly on t...