DSXchange

DataSet files do not need to be named with a ".ds" extension, they can have any name.
DataStage PX uses DataSets as faster, more efficient storage than sequential files.

What kind of a load are you doing in the Oracle stage? How many columns in your definition and is the total length 272 or 273?

I just reread your post and notice that the error message is one that refers to a source file stage. Are you 100% certain that this is referring to your output Oracle stage?

If you go into your favorite SQL tool and do a "desc {YourTable};" you will most likely see that the field is declared with a length of 273, but you only declared it as 272 in your job.

Your post confused me. Even after re-reading. To debug your problem, simplify the issue. Use 1 row of data, truncate the table and run the job with a reject link. Is the row inserted (check using your favorite tool) and if not, is anything written doen the reject link in the job or visible in the lo...

Did you put a reject link in to measure this? If not, you won't know. If the records are NOT being written to the table then they will go down the reject link

Rea - If you use a D conversion without specifying the order, the ICONV() function will make a best guess. If you have 4 digit years and alphabetic years then a format of "D4" will always convert correctly. If there is any ambiguity in the input date then the ICONV will most likely produce some inco...

That type of a range lookup will only work on Database objects. You need to use a sparse lookup and user-defined SQL in order to specify your BETWEEN condition.

Rea_Anthony - please write a test routine with a 'D4' conversion, then input your test dates and you will see that it does indeed cover most of your scenarios before saying that it doesn't work. Particularly try "2007-Jun-01" and "Jun-01-2007" and "01-jun-2007".

The commands you need to do this are in the BASIC Programmer's Guide and are OPEN , READ and CLOSE . Example code: OPEN '','MyHashedFileName' TO WorkFilePtr ELSE CALL DSLogFatal('Unable to open MyHashedFileName, status is "':STATUS():'".','') READ MyRecord FROM WorkFilePtr,...

One way is to execute the Db2 select clause from a command stage in a sequence, then pass the result as a parameter to a job. Sequences cannot read tables directly.

When you specify a simple "D4" conversion code the routine will make a best guess as to the format, but it is very poor practice as the computer has no idea if "2007-04-08" means August 4 or April 8; thus you will certainly get invalid conversions.

You need to code a reject link in a transform stage right before your oracle stage and capture your error.

If the content is still the same garbage then you have a record length issue you need to fix first.

As Ray has posted twice and I once in this thread, you need to match like-for-like. Your stream and reference lookup columns must be of the same datatype and/or format in order for the stage to work. Hence the questions asked above regarding your definitions.