DSXchange

A hashed file can have only one primary key (at a physical level), but it can have more than one column in that key. The total number of characters in all key columns may not exceed the number specified by MAXKEYLENGTH parameter in the uvconfig file. To perform a lookup against a hashed file, a key ...

DSGetLinkInfo() requires four arguments; a handle to the job (DSJ.ME may be used), the name of the stage, the name of the link and an integer specifying what information about the link is required to be returned. DSGetLinkInfo(hJob,Stage,Link,DSJ.LINKROWCOUNT) will return the number of rows passed a...

The Extract() function works with dynamic arrays to extract an element therefrom. Your function (Extract("dsxchange",1,2,2)) attempts to get the second sub-value from the second value of the multi-valued first field of a dynamic array but your first argument is not a dynamic array, merely a scalar s...

Do you have MKS Toolkit or some other UNIX emulator available? If so, you can use the command wc -l filename to get the line count.

No.

Type 25 files should only be used for indexes. They do not use a cache. Ever.

There are other utilities for examining Type 25 files, such as blook, but I am not going to pursue that course of inquiry, since it's totally irrelevant for DataStage.

Not totally true, but close enough. The way to do it is to read the file as containing a single VarChar column, parse it in a Transformer stage, and make the accept/reject decision while parsing.

"user defined", surely?

Yes, that's what it does.

Remove some CPUs, buy more licences, or set up a message handler to demote this particular message from warning to informational.

What about the control files (*.ds)? If you specify these, using either the GUI or the orchadmin command (orchadmin rm controlfile.ds), then the actual data files are automagically removed too.

Specify data type as Date when reading the Sequential File, with an appropriate date format string. That will convert the text string into a date, so that you need do nothing at all in whatever Oracle stage you are using.

Disabled. RCP may aid developer productivity, but leads to a situation where lineage analysis can not be performed.

I'm not sure that they know what they're talking about either. Hashed file lookups are the fastest around. If they're saying that your "code" needs to be optimized, ask them "what code?". You're working in a GUI environment, not a code environment. If they're so clever, why don't they do it themselv...

Your hashed file was created successfully but not as a 64-bit hashed file. I don't know why; your mkdbfile command looks OK. Can you perhaps try it with the -64BIT option in upper case? How did you execute the mkdbfile command? For example, was it through ExecSH?

Every job that runs runs as a "phantom" (DataStage terminology for "background process"). Each process records its start, its finish and any other unusual information in a file in the &PH& directory. Ordinarily any unusual events are transferred to the job log, but this may not be possible i...

It's not that IBM don't want to help you more quickly, it's just that its internal procedures are so cumbersome. Alas.