DSXchange

If it looks like an interview question, sounds like an interview question and quacks like an interview question...

Not the first time this one has been asked here, either.

Row limits for reject (and other) output links from a Transformer stage are set in the grid dialog where you set constraint expressions and identify the reject-handling link. (They are not set as run-time limits.)

DataStage Administrator should present you with a list of projects and wait until you have selected one. If this is not the case do contact your official support provider, because this is not correct behaviour.

Start > Control Panel > System (Advanced) (Environment)

You may need to re-boot for environment variable changes to take effect.

Not sure whether the bug's in the product or in the documentation. I work around this problem by analyzing a virtual table containing only the columns in which I am interested.

It is definitely NOT the same as migrating from version 7 to version 8. It is far simpler for a start. Are you intending to upgrade on the same machine, or install cleanly on a new server then migrate? The latter is recommended practice if you can manage it.

This operator can appear, for example, if you're using Sort Merge as a collector algorithm or if you've specified a Sort Funnel. Your score will tell you precisely where the Sort Merge operator has been used. This operator has reported to you that there is a reportable imbalance in the number of row...

That is not an error. That is an alert message. It informs you that the reject-handling link has been used, and how many rows passed along it.

Or, more simply:

Yes. Do the test in stage variables. Have three outputs from the Transformer stage, each constrained by the same expression. Use a Funnel stage downstream to bring them together if that's required.

The function behaves in the same way as the stage.

Merge stage consumes rows from its Update input link(s). That is, they are deleted from the "reference" data set as they are used.

I can't answer that directly unless you tell me how your Surrogate Key Generator stage is progammed - that is, whether it uses a state file or a database sequence. In any case it does not "know" anything - it simply delivers the next block of key values each time it is invoked. Surrogate k...

Even that is sub-optimal, because you re-evaluate Char(13) for every record. Better would be to evaluate Char(13) (and, possibly, Char(254)) when initializing stage variables, and not to derive those stage variables further. Then Convert(svFieldMark, svNewLine, InLink.TheString) to effect the conver...

No, what I am saying is that the first row into any cache is written out first.