DSXchange

Not really enough information. However, the problem is that one of your SQL statements in your job contains a number of parameter markers that is different from the number of columns defined on that link. You must find out where that is, and fix it.

What (exact) error message is logged?

Remove Duplicates stage gives the capability to preserve the first or last in each group.

Start with the main one, which occurs between the keyboard and the chair.

Yes, basically what we described. However, for that to work from generated SQL your table will need a SYNONYM defined for the column that has to appear twice. This is one of those cases where user-defined SQL is an easier answer.

Not "I came across a whitepaper", but please provide a link to the white paper. Otherwise we can't confirm what you claim. XE stands for "extended edition" (if you like "with the lot"). If you search here you should find a post where Vince McBurney explained the various editions - or it may be in on...

DataStage UTF-8 map and NLS_LANG setting American_America.UTF8 should be compatible. Can you check that there's an entry in the compatability file (I've gone blank on the name of this file!). Apart from that, check whether you need LANG_C also to be set, and/or try creating NLS_LANG in Adminstrator ...

I think so. Len() is probably the most efficient function in DataStage, because the length of a (variable length) string is stored as that string's prefix. You find the length even before you find the string!

Only earlier as far as version 5.0 - it's a different structure earlier than that. Version 8.x it's different again. Further, some of the things I referred to above are stored in RT_CONFIGnnn.

In server edition dates are stored as days since 1967-12-31, and times as seconds since midnight. You need to tear apart your timestamp into date and time portions, and convert your increment into days and seconds. svStartDay = Iconv(Field(InLink.TheTimestamp, " ", 1, 1),"...

What is the value of the NLS_LANG environment variable (used by Oracle)?

A better fit is to download the UniVerse 9.6 documentation.

UniVerse 9.6 is the last release actually to have been used as the DataStage engine. There are things in UniVerse 10 that are not in DataStage.

In my experience nearly all "unique requirements" of this nature are best solved with a custom routine. However, what you need can be done with (for example) stage variable expressions. To get the location of the first non-zero character, trim the leading zeroes and subtract the length of that from ...

Prevention. Verify within your job stream that no referential integrity issues occur.

If you can't get exact counts, how do you know that the counts you are getting are not the correct ones?!!

You can use the API or dsjob after the job has completed to get link row counts. But this uses the same mechanism that DSJobReport uses.