DSXchange

So IBM have pretty much admitted that this is a bug. Their 'solution' is to not use views of the XSD, and to just build the whole thing at once. In our case, this just isn't practical. We have around 60 datasets being used to form the XML, and having 60 H-Join steps in the hierarchical stage isn't r...

Still waiting on a response from IBM, but it seems that this error only occurs when the xsd/view to which we are mapping contains elements that are sub-types. For example, we have an XML element type 'Policy' with a number of columns which are inherited by all policy types. We then have a number of ...

Thanks for that, Ray. I was afraid that was the reply I was going to get.

We've now raised a ticket with IBM, so I'll report back here once we get an answer.

Hi all, I'm new to using the Hierarchical Data Stage, but I'm persevering with it and slowly I'm getting the hang of it, but I'm seeing an error with one of my jobs that has me tearing my hair out. Basically, I'm trying to create XML using the hierarchical stage, from around 20 datasets. DSs are all...

Finally found a solution for this, in case anybody else was having similar problems. Instead of using my custom SQL directly in the ODBC stage, I had to construct it as a text string in a transformer first, then pass that string as a bind variable to be executed by the EXEC() function in the ODBC st...

Hi, I'm having a problem getting bind variables to work correctly in an ODBC Connector stage which is performing an update to a SQL Server 2012 table. The SQL I'm trying to get it to execute is : update TBL_BlobData set Data=(select * from OPENROWSET(Bulk 'orchestrate.MediaFilePath' ,SINGLE_BLOB) a)...

Just a short cautionary tale. If anyone else does this, use some common sense and make your changes to a copy of your .dsx file. Naming no names, but I once worked with a guy who ignored the development standards and instead of naming one of his job parameters 'db2_password_str_parm' like the rest o...

I've seen a similar issue before when reading sequential files where the last two columns are 'null'.

Can you try setting the datatype of the problematic columns to varchar?

That was the workaround I found.

Have you checked the permissions of the file / directory to which you're trying to write?

nani0907 wrote:here is the command line prompt to get list of JObs.

dsjob -ljobs <Project NAME>

That's a great tip. Thanks Nani.

If you're familiar with the Universe shell, you can query the DS_JOBS table to find what you need.

You'd simply need to use:

SELECT NAME FROM DS_JOBS;

In your output links, is ColB defined as 'Nullable'?

... but why go to the trouble of Googling something when you're here, Arnd?

You can find plenty of detail on how to use this stage in 'The Parallel Job Developer Guide' in the Documentation folder in your Information Server installation folder. If you're still struggling, post back with specific questions about what problems you're experiencing, and I'm sure someone will be...

To split the data into columns, you can read the file as if it contained only one column, with all data for each record in that single column. You can then pass this into a transform stage and use the field function (with the comma character as delimiter) in 26 stage variables to split the data into...