DSXchange

I'm not a place where I can check the functions, but I think this particular one and it's functionality is missing in PX. If you just have lowercase characters to worry about, you can use CONVERT('abcdefghijklmnopqrstuvwyz,'',In.Column) Add uppercase and punctuation to that list if necessary...

Since you seem to be overlooking what Craig and I posted, you will have to wait to see if Ray responds.

The "EVAL" performs a function and is compiled at the time the SELECT is issued, the two first lines are the output of that compilations of the 2 EVALs in the command. @RECORD<14,1> means "get the value of the 14th field in each record of the select.

A text file won't contain NULL values and you cannot trim a CHAR column. Either set a column default value or check for empty using a derivation such as 'IF Trim(In.Column)=""' which will create a temporary VarChar as the result of the trim() function. Or check for the correct number of spaces, i.e....

The number of input columns needs to match the number of columns declared in the SQL, so you need to remove the ADDR_KEY column from you input tabs and then shift the relative columns numbers over by 1, i.e. add a :1 and remove the :23

I assume that you mean incremental backups in a project, where a subsequent backup only saves those jobs which have changed or been added since the last backup. The mechanisms that support automated backups of DataStage unfortunately do not have such a capability. It is possible, through use of the ...

Hmmm... could be in the write to sequential. Please put in a constraint "1=2" in your transform stage, recompile and re-run to see if the error remains the same or goes away. Also, how much data / how many rows are you writing?

From a sequence all you need is the Job Activity stage, and there is no difference there in how the jobs are called.

Yes, many active stages like the transform and copy stages support both input and output links. The passive stage functionality you are referring to with server jobs does not exist in Parallel jobs. The "other" link in these stages is, where allowed, always a reject link in PX jobs.

Mark_e - just remember to add a dummy line to the data so that the last and final group change is detected, otherwise your output will be missing one or more lines.

Since "convert" is already a function, perhaps choosing another name might help.

<sigh> The DataStage value is an empty string. DataStage doesn't care about data types so accepts that value. When that is being written to your Database it won't accept that invalid date, hence it uses NULL. You need to fix your derivation so that it results in a valid date in the correct format. Y...

I read this one while it was still in the PX forum. It has migrated through osmosis....

Craig - I based my post on that statement as well, since you cannot compile a transform stage without a correctly installed C++ compiler in PX jobs. If the simple job had a transform and compiled then it must have been a server one.

Are you talking about a NULL or an empty string? The former, in Server, can only occur if the input is already NULL!