DSXchange

I am confused as to whether your problem is in the Job Sequence and getting the 2 values or whether it is in the PX job after passing those two values?

PX makes that kind of debugging difficult. Do you know which particular conversion is causing it? If not, you would have to set APT_DISABLE_COMBINATION to find out which transform stage.

Yes, I read the title part and that is why I put my post in. I didn't realize you were going from non to nullable, in which case you can do this in just about any stage.

You cannot link directly to ODBC on UNIX without {buying and} installing 3rd parts software.

Naah, that isn't the case when it comes to opinions. And I always favor bounded over unbounded strings but when it comes to DataSets with large volumes they save both space and time. Using the fixed-length max size makes indexing faster, but the cost in space is horrendous with strings that are "spa...

Do it in a transform stage, but you need to ensure that the new, non-nullable, column doesn't contain any nulls. You can use several null-handling functions to do this.

Actually, I stated the exact opposite regarding storage size of unbounded VarChar() fields for datasets.

That's the first mention of SAP that I saw in this thread. I thought you wanted Excel to keep the formatting and that is what I played with. Can't help you with LSMW.

In your job sequence you have that information available to you. Each Job Activity has 3 User Activity variables associated with it, the $JobName, $JobStatus and $UserStatus. You can use those in the sequence, for instance in a User Variables Activity, to set a string value with the job name of the ...

I tried it and you are correct, then I played around with the CSV format and found that



does import correctly. Could you change your job to write out this field format?

If you use unbounded VarChar() columns you will use a lot less space if your strings are nowhere near to being at their maxiumum length. for example, with a single node if you create a DataSet with 1 million rows declared as varchar(1024) and fill them with "hello world" the file size will be 984.38...

Can you not explicit declare them as VarChar() columns, then they get quoted in the .csv file.

What kind of contents did they have?

Raftsman - for a while (during my wait for my US visa) I was working out of the Toronto office and we had a number of early adopters including Shoppers Drug Mart and I thought that the DS market was doing quite well. Then again, it has been a couple of years... Or are you referring to the Francophon...

Could it be the default file name for a sequential file output? Have you tried "file <name>" to see what type it is or looked into it via a command such as "head -10 <name>"?