DSXchange

I thought there was some way in PX to have the Sequential File stage return only the fully qualified pathname of the file found. Rather than trying to make it do the folder trick of 'reading the entire file into one record' - which while it may work sometimes will never work 100% of the time - see ...

Exactly what we've found. The way DataStage 'formats' the output sucks. It really shouldn't matter to anyone processing the file automatically, but when people are looking at it you need some kind of formatting or 'pretty printing' to make that feasible. Or you need to get them a copy of XML Spy and...

And by 'generic' stage, I assume they mean... transformer?

In 'real time mode' they can be set to always run, so no startup time per se. Otherwise, you have the 'normal' startup time when a request is made and the jobs aren't running, and then you can shut them back down after a user-defined period of inactivity.

It happens like how regularly? Monthly? Daily? And no chance of Change Data Capture being setup... or a trigger on the table(s) to capture deletes? I'm not sure a wholesale delete of X days worth of data and then a reload of that is the answer to all your woes, but it certainly can be done. DSguru2B...

Exactly... parse the columns out and then pivot. Don't think you need to 'delete' the original record, just don't pass it on.

Very nice. :D

The SOA Edition - formerly known as 'RTI' or Real Time Interface. Sure sounds to me that's what you are describing...

Closest I could find on the IBM site:

http://www-306.ibm.com/software/data/in ... tures.html

Not sure where the 'UV skillz' come into the picture. We're doing big complex XML with large numbers of fields with no 'UV' (or custom BASIC code) in the mix.

Regardless, good you have an alternate approach that works for you.

Seems like it is time to call Support and ask them.

Have you read through the XMLPACK_20_Designer.pdf in your Docs directory? There are quite a number of tips there on generating complex XML output, including how to use custom stylesheets. I've found very few situtations (only one actually, so far) that required a CSS. For your problem, I would have...

You need to do this all the time in a job? Or just one time?

But then come back and reveal all.

Those are XPath Expressions not 'tags' that generate the XML output in your job. And you get the 'name = "xxx"' elements because of the use of the '@'. I'm not positive you can generate output like that without doing your own 'tag building' or XML creation. What I would suggest you try is to take a ...

Ok... more questions. When you say 'changed the join in the transformer to XXX' do you mean that, rather than simply put the input date field into the key expression of the reference lookup, you are using those transforms there? Shouldn't be needed as they should both work 'as is'. If both of your s...