DSXchange

You can also run Progress utilities to dump data into flat file formats (even initiating these utilities from DataStage) and process those files. This approach allows hierarchical structures to be unpacked.

Check the purge settings for that particular job (choose Clear Log from the Job menu to see them).

In the GUI tool choose Save As to create a new configuration file name (which ends in ".apt") and edit that. Don't forget to click Check to make sure that there are no errors or warnings.

There is an error in the Parallel Job Developers Guide that asserts that the NullToValue() function is available in the Modify stage. It is not. The function that is available in the Modify stage, as noted above, is handle_null().

A quick fix is to set the value of PWD to "." (the current directory).

How about using a Remove Duplicates stage? Or even a unique Sort?

saikir wrote:The existing job accepts only Windows line termination.

But it could easily be changed (stage properties of Sequential File stage).

It will never be possible.

In Version 8 the message changes to something like "opened by user xyz, would you like to open in read-only mode?". So, while it's possible that more than one copy might be open, only one of these (the first-opened) is editable.

1. 32-bit. 2. It's quicker to truncate. And the permissions required are different. 3. When you choose DB2 partitioning algorithm there is a tool to the right of the drop-down list, which opens a dialog in which you can name the partition table. 4. You don't need to - the DB2 partitioning algorithm ...

An inner join on an Entire-partitioned constant (the solution I posted earlier) will generate the required Cartesian product.

Does 0x04 (Ctrl-D) appear anywhere in your data? UNIX (and the STREAMS I/O module in C) will recognize this as an end-of-file marker.

Add a rejects link to the Sequential File stage into a Peek stage. Look in the entry logged by the Peek stage at the raw form of the rejected records.

Please post the first couple of lines from each of the input files. Please also verify that these are true text files, not hashed files.

Perhaps two Index() functions as the arguments for substringing; one for the open tag and one for the close tag, would be a more general solution?

Use the Hashed File Calculator; it will take storage overheads into account. Even so, with one million records with 12 bytes of data, that's still only 26MB or so. That is well within the range that a hashed file can manage.