DSXchange

Arbitrary levels of recursion aren't easy to do in any ETL tool and don't really belong there (in the same way that RANK doesn't). These really are display functions, not transformation functions.

Welcome aboard.

What exactly IS "%mmmm" ? I don't understand what you mean by it.

Subtract the time when the command is issued from the time that the command finishes, and you will have your answer.

1. That's the kind of thing I do for a living. See this page for example. 2. It's rarely a safe assumption that any answer relating to parallel jobs applies to server jobs. 3. Where's the business case? IBM don't do anything that won't turn a dollar. 4. No more than in earlier versions. Even the pos...

Nah. All arguments are passed by reference - that is, the address of the string in memory is all that's passed to the routine. That was, of course, a slight lie. What is actually passed is a DATUM, an internal, 16-byte structure that includes the address and also other information about the string, ...

EBCDIC() is an internal function in the language, doesn't need DEFFUN.

WRITEBLK is a statement, not a function. You write precisely the byte values that are in the variable. No line terminator character is written; you must organize that yourself.

There's no OPENBIN, you just use OPENSEQ as normal but you write blocks of bytes using WRITEBLK rather than writing lines using WRITESEQ. You still need to close the file using CLOSESEQ.

I'm not really sure what you're asking here. Yes you can create a Connector for SQL Server, and you can load that Connector with a Parameter Set containing all the parameters you require. But "like we used to do?" - this is new technology in version 8.0.

You need to be attached to the project directory to execute DS.REINDEX, not to the engine's bin directory. You also need to have executed the dsenv script before executing uvsh.

It is easy enough to write such a function. Probably many have done so.

Resist stupid requrements!

Let the metadata do it for you.
If you must, define a fifth field (VarChar). If it is populated, then the line has too many fields.
Too few is a harder problem if field #4 can legitimately be empty.

Talk with your Oracle DBA about the maximum number of sessions permitted from the one client. In this case the Oracle client is your DataStage job (your DataStage server and any other hosts named in the configuration file).

In server jobs and job sequence expressions, yes. In parallel jobs no - you need to convert to any numeric data type before arithmetic can be performed.

Ask yourself two questions.

What is the ASCII/Unicode codepoint for "."?

How are strings of characters compared by a computer?

The cause is purely and solely having to wait for Oracle to respond.

You can get way more speed using a state file, which you could initialize from your Oracle sequence then update the sequence once the job had processed all the rows.