DSXchange

Only good if there are no "." characters in your data that you want to keep (decimal points?).

Anything but a hashed file stage can accommodate your requirement, but you'll need to us user-defined SQL (for the BETWEEN condition) and specify your search keys carefully.

Is there a 'simple' option to write directly into a table from a routine?

Yes. Write to a sequential file and invoke the table's bulk loader through DSExecute.

Any plug-in installed with the package installer installs only into the projects selected when installing. It may be that the QS plug-in doesn't let you select more than one project but I would think that unusual. Probably there was only one project when it was originally installed.

If DataStage isn't running, and nothing else that uses shared memory segments is running, then it's possible that ipcs -mop will return no information.
Try starting DataStage before running the command. Also try being less restrictive, say ipcs -m or ipcs alone.

Looks like there's something about the table name that the ODBC driver or JET doesn't like. Can you post the generated SQL?

You will need to create a custom routine. There is no intrinsic function for this particular task.

Open cmd, then type AT. This will report which jobs are waiting to execute.

You may have hit the maximum size of your hashed file write cache, which would definitely result in a major slowdown. Setting MINIMUM.MODULUS, as Arnd suggests, pre-allocates the disk space, but does not ameliorate the fact that writes into this space are to "random" locations. That's why memory is ...

The Lookup stage requires a minimum of two input links - the stream input and one or more reference inputs (against which the key-based lookups are performed). On the other end of a reference input there can be a virtual Data Set, a Lookup File Set stage, or a DB2 or Oracle stage.

If the RT_LOGnnn hashed file has become corrupted (for example by attempting to exceed 2GB or file system full), you will continue to get this error until the log is cleared or repaired.

Welcome aboard! :D Your analysis is sound. However, there is no documented limit on the complexity of sorting. Get in touch with your support provider; the query sort engine can be tuned, but it's a delicate process (you alter parameters called QS... in uvconfig - read the dire warnings posted there...

Use DS_CONNECT rather than CONNECT. This has authorization for the branded ODBC drivers built in.

No-one suggested updating the CATEGORY column in the underlying Repository tables. This is good, people are learning to avoid hacking the Repository, whose structure is about to change radically.

Isn't the delimiter character a property of the stage/link?