DSXchange

Is the licence on the C++ compiler in use when you try to compile these jobs? Does it happen if the job does not contain a Transformer stage?

Welcome aboard. The short answer is yes. A Left Outer Join is the whole of the "left" circle (set), including the intersection with the "right" circle, and nothing else from the "right" circle. This does not adequately show the substitution of NULL for column values tha...

Not seen this one, but I skipped FP1. Check out whether it's a known issue fixed in 9.1.2.

Don't send duplicate rows. This might be because the same key has been processed in different partitions. The Remove Duplicates stage only removes duplicates per partition; therefore your data should be partitioned using a key-based algorithm on the Remove Duplicates key ahead of that stage.

It varies. Use OSH_PRINT_SCHEMAS environment variable to have DataStage report on schemas used, but it will not show a length for an unbounded VarChar.

That's where I was heading when I asked you for your xpath expressions. Ernie pre-empted me. But that's OK - the more people answering questions well, the better.

Describe the XPath expressions and other settings you have used in the XML Input stage.

Are there any settings in the second Connector that might cause the After-SQL command to be skipped?

Is the after-SQL properly formed?

Or just run the SETFILE command from an Administrator client Command window in the project.

You can do this by using a UniVerse (UV) stage.

For that to work, there must exist a VOC pointer to the hashed file.

Try disabling tsort operator insertion, either using an explicit Sort stage (set to "Don't sort, already sorted") or the environment variable.

No computer can store float data type accurately 100% of the time. There are IEEE papers out there explaining why. You could try changing the data type to DECIMAL(38,2) or using explicit rounding functions.

That's driven by link execution order. Left with first intermediate -> result1, then result1 with second intermediate -> result2, then result2 with Right.

Not in Data Set creation.

But more alerts are generated, that were basically ignored in version 7.1.

Hard to tell. More stages may result in more processes, which may help throughput (assuming operator combination to be disabled and a server that's not overloaded).