DSXchange

Sort stage is primarily there as a marketing exercise; once upon a time Informatica beat up DataStage by observing that DataStage lacked a sort capability. So they put one in. UNIX sort is faster than the DataStage one. CoSort and SyncSort are faster still (faster, too, at emptying your budget).

It's not officially "version 8" yet. It's called "the Hawk release" and it's just gone into the first of three phases of beta testing. Scheduled general availability is September 2005. The main changes are ("Sorcerer") an active dashboard giving access from a single point to all Ascential products, ...

Probably THE most important rule for Link Collector and Link Partitioner stages is identical metadata on all input and output links. You have encountered one of the things that often traps folks new to shared containers. I make it a practice to document the columns being processed in a shared contai...

It's more likely that NLS is not enabled on the server.

Even though you install NLS, there's another step required to enable it. That step involves changing the NLSMODE parameter in uvconfig to 1, then regenerating and restarting DataStage.

Status 99 also covers the situation "status file has been cleared". This could have been done even if the job is not compiled, for example if it had been previously compiled but has been changed and saved but not yet re-compiled.

Your after-job routine can also interrogate the job to determine information about parameters - not just value but also type, default value, and so on.

The function you need is DSGetParamInfo(). Use DSJ.ME as the job handle (this means "the current job").

Whatever you tell DataStage (via SET.TERM.TYPE) must match the terminal type that your emulator (PuTTY) is emulating. For example, if PuTTY is configured to be emulating vt220, then is the command that you need to issue.

The same is true if you use Link Partitioner and Link Collector stages, with the extra consideration that each also creates a process for itself.

Shared hashed files require a reconfiguration of DataStage. For details see dsdskche.pdf in the DataStage manuals.

You can trigger the abort condition by limiting the rows on that output link (rightmost column in the Constraints grid). However, I do not think that that number can be a job parameter. To intercept and abort you could (for example) test @OUTROWNUM on that link and fire a routine when it exceeds you...

You could also store the value in the job's user status area, and retrieve it from there in TransformerB.
However, under the covers this does a write to disk anyway, so there's no gain (and there is some pain) in using this method.

Do a search for "vertical pivot". It can be accomplished with an intermediate passive stage, even though this runs counter to the PX design metaphor of doing everything in memory.

You could also create a custom stage to perform a vertical pivot.

Use a hex editor to determine what the trailing characters are. It will then be easier to control DataStage to avoid writing them.

The only thing I can imagine is a quite complex expression explicitly testing each character position, with some superimposed logic that tests ordering (for example no embedded spaces). It'll be ugly, since it has to cover all possibilities, but should be do-able in DS390. Plan it all out on paper, ...

DataStage does not have any problem with the character "0". I'd guess that you don't have this in your data. Use a hex editor to determine exactly what you do have. The ASCII code for "0" is 48 (which is 0x30).