DSXchange

You can also get it from the Reporting Assistant. The SQL is a property of the input or output link.

Define a table definition consisting of four columns. The fourth column is the remainder of the record; figure out its width from the metadata. Or, if it's variable length as you say, then there must be a line terminator, and you declare the data type as VarChar. You can read the file using this tab...

You can execute CREATE.INDEX and BUILD.INDEX via the before/after subroutine ExecTCL; coordination of these after the hashed file is created will be problematic. Or you can create your own before/after subroutine to do it all - delete and create the hashed file and its indexes. If you do this you ca...

We share here. Please post the solution.

/dev is a system directory; it does not exist as a subdirectory on UNIX. The story is different on Windows, where there is a NUL file in every folder.

Completed months (always integer) or partial months? For example, what would be your expected result if the two dates were 2004-01-02 and 2005-03-15 ? Your choices are 14 or 14.4. A routine is needed to handle the situation where the day number of the first month is larger than the last day of the s...

No. It's not an Ascential issue.

The issue is that you were relying on what was, essentially, a loophole in enforcement of data types. That loophole has now been closed.

It is, and always should have been, a requirement that your data must conform to the metadata that describe them.

What everyone is saying here is that you appear not to have a specification for source-to-target mapping. The easiest solution is cat fileA fileB fileC > fileD which will place all rows from all input files into one output file. If that's all you want, you don't need DataStage. So: what are the meta...

It's not too expensive, since you can't guarantee that you're violating primary key constraints without doing it! And it's not expensive. Simply pre-load a hashed file with just the primary key values from the target table, and do a lookup against the read-cached hashed file. This will determine whe...

If you need the value in a downstream stage, just create another column on the link(s) leading from this stage to that.

This is by far the most efficient mechanism for passing a value from one stage to another.

The question about why jobs that have their processes killed has been answered a number of times, try a search. In short, they don't get a chance to update the RT_STATUSnn table associated with the job (which you look at through Director). Never use kill -9 on DataStage processes. Prefer something m...

Grouping and summing are the preserve of the Aggregator stage (though it can be done more clumsily in a Transformer stage). Many other aggregate functions are available but First and Last are not among them. You therefore have two options; to use a Transformer stage and perform your change detection...

No.

DataStage is metadata-driven, and does not allow column names to be job parameter references. In any stage type.

You don't need administrator privileges, but you do need to use the Administrator client. When on the Projects tab there is an NLS button.

... and, of course, there's a version for Tru64. But I don't believe there yet exists a version for any 64-bit Windows operating system.

As Roy suggests, get and official answer. Then please post it back here.

That's because any decimal[n,k] can automatically be "promoted" to any decimal[n,x], so the scale has no effect in your expression. My point was that you needed to have something set for scale.

Surely you're not old enough to remember Laugh In?