DSXchange

The empty value is in the KEY, you can identify the column of a compound key only by splitting the columns individually, for instance by another output stream to a sequential file that splits out each individual column.

The CONVERT function is character based. You can use the EREPLACE function to do string based replacements and call it up twice, once for the "M" to "10" and another time for the "F" to "20" replacements.

You still should load you hashed file to memory and lookup each row, but instead of using a reject link just put a RefLink.NOTFOUND constraint on your output.

Are you certain that you have recompiled, perhaps could ensure a clean compile by using the "force" recompilation in your test environment.

The log message doesn't affect the number of rows being read, just the number of concurrent files that the process has open. Your error must be coming from elsewhere. What do you mean by "with pre-defined limit of 100 records"? Do you set a limit yourself somewhere? The number 100 is too round to be...

The disk sizings for similar jobs can vary greatly. If you read a database and pre-sort the data on the select then a typical job might not need any interim storage even for gigabytes of data. If the incoming data isn't sorted then any aggregation or similar operation might trigger a temporary stora...

Naming conventions are not absolute. Most organizations that use DataStage already having general naming conventions in place and adapt those so that they fit DataStage. The good and comprehensive naming conventions that I've seen or worked upon are all proprietary to their respective companies. A g...

Yes, but it depends in which type of object you wish to do this. If you want to do changes in the Designer you can do them along the lines you put, but there is no standard interface to do this for all objects in the repository (although you can work with the repository directly, although it is neit...

Umm, even with shared containers you don't need to use IPC stages if you have inter-process row buffering enabled.

Have you changed the IPC stage attributes from the default values? Are all 5 containers you use in your job from the same shared container?

The Hawk repository is no longer the DataStage flavor of UniVerse. Informatica has never used UniVerse files and cannot access them (without the 3rd party ODBC drivers).

I am somewhat confused by these continued posts on this subject. On Friday I mentioned why it won't work, and Ray stated it again this morning.

Are you using IPC stages or do you have the job attributes set to use inter-process row buffering? If the second, what happens when you turn it off? Also, are these shared containers?

As Ray has said, the @NULL is the correct value to use. Are you certain that the column is nullable in the database? What is your rejection error that you are getting?

Deeps - I have had that happen to me as well. It reminds me of a post in an aircraft log where the pilot wrote "Evidence of oil leakage in starboard engine number 2 found" and the maintenance engineer repleid "evidence removed" Seriously, there is nothing anyone can do with that statement. Could you...

Gillu, until you know what your data is you will not find a solution and nobody here will be able to help you. NLS and character sets are irrelevant to comp-3 data since you need to make sure you don't apply any mapping at all to this binary data. Please use your favorite editor which is capable of ...