DSXchange

kcbland wrote:There is no need on Server jobs to cast any value. Server handles everything has a text string. If you perform math on a value, it will automatically cast it to a numeric and do the math.

He meant "as". Server handles everything as a text string.

See, for example, http://www.dsxchange.com/viewtopic.php?t=86343

If you're talking DataStage job parameters, simply edit the job properties and create the parameters. Make the pathname parameter's type "Parameter" so that, if the job is invoked manually, a browser is made available to the operator. You can invoke your script with DataStage job parameters on its c...

The best way to do this is with an edit macro. Unless you are very comfortable with the Universe editor then do not even think about this. But a friend of mine Peter Lai did this and saved a lot of time. It could work on any string which is clearly defined without any worry about changing the wrong...

I make it a habit to remove the qualifications on table names immediately after importing table definitions. This means that they can be loaded into column collections unqualified. The downside is that, where I want user-defined SQL to extract from multiple tables, I have to add table aliases/qualif...

As the Administrator (a user with DBA privilege), either from the client window in Administrator client or via a telnet session to the project: GRANT CONNECT TO cmachado; GRANT SELECT,INSERT,UPDATE,DELETE ON tablename TO cmachado; GRANT CONNECT TO FOCUS\cmachado; GRANT SELECT,INSERT,UPDATE,DELETE ON...

My current client is upgrading from 5.2 to 7.0.x sometime in the next month or two (depends mainly on delivery of hardware). Based on discussions in the local user group (in Australia) and elsewhere, we think that 7.0 is sufficiently stable (server jobs only) for our requirements.

Purely from the point of view of not having to restart the server every month it's worth doing, imho. UNIX is far more scalable than Windows, and can therefore handle much greater volumes of data. Also, PX is not yet available for Windows. Some UNIX systems use mutex locks, others don't. If you're c...

Direct is possible, and others have done it already.
Search the forum for tips, tricks, caveats and gotchas.

One that jumps out immediately is that LD_LIBRARY_PATH does not include $ORACLE_HOME/lib or any other Oracle library. ORACLE_HOME=$ORACLE_HOME doesn't achieve much! PATH doesn't include $ORACLE_HOME/bin or any of the other Oracle executable-containing directories. Chances are, that TNSNAMES.ORA may ...

Please post full detail of the message logged. In particular, we need to be certain that you've correctly identified the stage that generated the error. "ipc" stands for "inter process communication". It may be that the ipc file referred to has something to do with row buffering. Can you also advise...

A Hashed File stage is purely for key-based "equals" lookup. You provide a value via the reference key expression, it finds the record for which that value is the key. If you lie about which column is the key column, more than one row may be selected but the Hashed File stage has no way of generatin...

NULL is "unknown" value. Pretty much any operation that you perform on NULL returns NULL. So, if you try to ask "is this equal to something whose value you don't know", it can only answer "don't know". There is one other way to do what you're interested in doing, a technique called a "substitution c...

Extract(Teradata) ---> Transformer ---> Load(Teradata) Use user-defined SQL in the Extract stage to extract from the three joined tables. This is MUCH easier - especially since you already have the SQL - than loading hashed files from the same database instance. The Load stage can u...

Normal courtesy on finding a solution is to post that solution to the forum, so that others can benefit too.

Presumably you either made the field larger, or used a substring technique to limit the number of characters that could be output.