DSXchange

In addition to what the others have suggested, please pay attention to maintaining professional standards of English. Not everyone reading posts here has English as a first language.

There is no Join stage in server jobs. A lookup in a Transformer stage naturally performs a left outer join. It can be proven that, by exchanging the left and right links, any right outer join can be constructed as a left outer join. This answer ignores issues of "performance", but so does the quest...

Derive the output columns directly from the stage variables, but only in the rows where the stage variables change. For example

You might start by looking for anything that might cause a memory exception, such as delivering NULL to a function that can not handle NULL, or division by zero, or attempting to shoehorn too large a string into a bounded string column.

Use two consecutive Transformer stages, one for each test.

Or use two output links, one for genuine writes, the other for exceptions.

In version 8 there is an Errors and Messages manual. I have not checked to determine whether QualityStage messages are included.

No. The accented European characters are handled quite happily using the "8-bit ASCII" (ISO8859-1) single-byte encoding used by DataStage without NLS enabled.

That's because some int64 numbers have more than 10 digits. DataStage feels that you need to be alerted to the possibility that you may lose some data.

If it's character-by-character the most efficient function is Convert().

Convert("AYXM","1234",InLink.TheChar)

You might then need to convert to integer of some kind.

Ah, that's rather a different problem. It is solved using a "fork join" technique - split your data into two streams using a Copy stage, run one stream through the Aggregator stage, and re-unite them using a Join stage. The join key and the grouping key are probably the same. ------> Copy stage ----...

File Set descriptor files are pure text. You might try editing one. But take a backup first, just in case. Things only go wrong if you don't have a backup.

However, this is not true for Lookup File Set or Data Set descriptor files. Don't ever try editing these!

You don't need to use a user variable. (Neither does U.)
The user status is available to any downstream activity in the job sequence as an activity variable from the (server) Job activity that set the value in the user status area.

You need to configure the Oracle client software. Ask your DBA for assistance. In particular, the tnsnames.ora file needs to be set up for all connections to Oracle that you might need to make. You need to configure ODBC data source names to connect to Oracle instances. This is done by editing the ....

You need to bring all columns into the Transformer stage then come up with some kind of merge strategy. For example If IsNull(LEFTcol) Then RIGHTcol Else LEFTcol Be careful to get this the right way around to handle the situation where neither of the columns is null - that is, that an inner ...

Include CAST functions in your SQL (the derivation columns) to convert to VarChar data type during the extraction phase. This will obviate the warning messages.