@INROWNUM and @PARTITIONNUM

bi_fujitsu · Post by **bi_fujitsu** » Fri Sep 28, 2007 1:04 am

We have a scenario, wherein we have 8-9 source datasets. After each dataset we have a transformer in which table_name is hardcoded. this transformer has only one column--table_name.
We are checking that if any of the dataset is having data, then that particular table_name (from transformer) is being written to the target dataset. The constraint in the transformer is '@INROWNUM=1 and @PARTITIONNUM=0'. We do have 2-node configuration file.
The problem here is, some of the datasets are working fine, but in 1-2 datasets, even if they have got data, nothing is going beyond transformer.
If anyone came across similar problem, please respond...

ray.wurlod · Post by **ray.wurlod** » Fri Sep 28, 2007 1:56 am

What a strange design! What the constraint means is that, if the table name is allocated to partition #1, it will never make it to the output. I suspect that's not what you want.

Forget DataStage for the moment.

Write out a specification (in English) describing your source data, what the target needs to look like, and the business rules that determine how rows get from source to target.

bi_fujitsu · Post by **bi_fujitsu** » Fri Sep 28, 2007 2:22 am

The scenario is, if any data exist in any of the datasets, the corresponding hardcoded value (in transformer) has to be captured in the target file. This target file will be used to run the job having that hardcoded value in its name.

The above design is fixed across various projects and is working for other projects.

ray.wurlod · Post by **ray.wurlod** » Fri Sep 28, 2007 5:54 am

So what's different about this project?