DSXchange

Start by making ECODE all Unicode or all not Unicode through your job design.

Then take a look at the "bad" file produced by sqlldr.

There are various vendors of training. Use the internet to find one or more near where you are. IBM does offer "e-Learning" but I have not checked for some time whether the Information Server suite of products' training is included.

Please be more specific about what you mean by a Value File. It's not a term that I understand a single meaning for. In what way do you intend to "generate"? And what has to be generated?

The email part is easy now that the Notification Activity supports attachments.

Don't encounter them. Prevent them. Recompile everything as soon as you import.

And make sure that you properly migrate parameter sets.

It should include the parent directory of liborchoraclent.so (or .dll or whatever). As well as lots of other things. As noted in your other post, this is really the shared library search list, rather than the command search list.

Where is liborchoraclent.so on your system? Is its parent directory in your shared library search path (LD_LIBRARY_PATH probably)?

What are your array size and transaction (rows/commit) size settings? If the latter is 0, then Oracle is building and building and building a huge transaction in the redo tablespace (rollback segment, if you prefer). The larger this gets, the slower things tend to be. The solution to that is to comm...

For a constant format, you can use pattern matching. For example a timestamp in ISO 8601 format with no fractional seconds: InLink.TheString Matches "4N'-'2N'-'2N' '2N':'2N':'2N" Or, to be a bit more generic about the delimiters: InLink.TheString Matches "4N1X2N1X2N1X2N1X2N1X2N" ...

It converts any field marks in the $Counter variable to zero-length strings. When DataStage reads (natively) a text file from the operating system it converts line terminators to field marks, so that the read results are independent of whether UNIX or Windows is the operating system. So, in effect, ...

Why not dump them into a pair of temporary tables (parent and child records) then use an export query with a join to populate the text file?

Any kind of outer join can legitimately return null when there's no key on the probe input corresponding to the current key on the driver input. The difference is likely to be in your data rather than anywhere else. Be very sure that your data are partitioned and sorted correctly (and that APT_NO_SO...

The reason that the conversion is happening is simply that you have NLS enabled.

Category in which branch?

At version 7 the answer is yes, but don't get into the habit because it's not nearly so easy once you move to version 8.

The complaint is that your (Brazilian) Portuguese NLS locale does not have a collate category - that is, it lacks the rules for how to sort characters. Get your support provider to verify that you have a complete installation of this locale.

No. As I mentioned earlier, the Index() function I mentioned is used in a Transformer stage - to do what you're trying to write a routine to do.