DSXchange

Line terminators are mapped to @FM. In your case you correctly (originally) specified comma as the delimiter.

The problem is probably with the line terminators in your file. Can you please share how you are implementing "read each line of a file at a time"?

Why not install the 7.5.2 client alongside 7.5.3 and switch between them as appropriate?

Does your file have an empty last line?

Sounds like you're in need of some training. Why not work through the QualityStage tutorial, which ships with the software and has its own manual? Cutoffs determine thresholds at which the composite weights calculated for records indicate a match, a non-match, or a "too close to call". Wei...

Too quickly, it seems.


The original error did seem to suggest that it was trying to read the column heading, since the first tab appears at offset 15.

In the first data record, the first tab appears at offset 2.

Did the error message actually mention a row number?

I trust that you have configured the stage to ignore the first line as containing column headings?

What happens if you set Array Size to 1 ?

See also this article.

When you define a blocking field in a match pass specification, you can specify whether that field is character or numeric. You have left the default (character) when you should have changed to numeric.

They do not propagate automatically. You have to open each Job activity and replace the parameter references with references to the new parameter names.

You use exactly the same command that you would use to run the script from a CMD shell.

I'd still look at getting more memory and using local lookup.

Sparse lookup is necessarily slow. Think about what it has to do.

Example: customer dimension. Name is Type 2, telephone number is Type 1. You never want to call people using their old phone number, but you do need to know previous name information. (This example may not be pertinent to a telco!)

There's no limit till you start running out of memory. However, every byte of memory you use for cached hashed files is marked memory that is not available for any other purpose. So the more you use, the less you have for other tasks. The whole thing's a trade-off - supply and demand.

At first blush I would presume that you can use ISM (or istool) to export from one repository and import into another without the need for DataStage to be licensed on the receiving server. That said, do check with IBM for a definitive answer.