DSXchange

Should mean the job aborted and thus failed to clean them up. If they are not currently in use and no longer needed, you should be able to simply delete them. Or you may need to use to use something like "orchadmin" if they have a descriptor file as well like a normal dataset. Others will ...

Workaround? No. You just need to learn how to deal with it. Did you try what qt_ky posted?

My source Oracle 9i table column definition defined as Number and Target Oracle 12C is also same. Not so much confusion as a lack of pertinent information. :wink: And just to verify again, your targets are both unbound NUMBER fields, correct? No precision, no scale. Those are (basically) stored as ...

Ah... so this works in the old XML stage but not in the new one? Sounds like a patch is in your future.

ps. Also moved you over into the proper forum.

We tried changing the 'data element' on the column to XML to tell the stage that the data was already encoded, but that did not work. I'd be curious what "did not work" about it as that IS the solution, that's exactly what that data element is there for. Have you involved support? Perhaps...

Possibly... as a test, remove them from the source and see. Or do a simple insert test in a tool like Toad to confirm you have the proper format mask for the TO_DATE function and your data.

I was assuming it was the error where Oracle will 'truncate' a unbound number [meaning NUMBER rather than NUMBER(20) let's say] during a fetch operation but that's another ORA error. The posted one is: ORA-01457: converting column overflows decimal datatype So also curious what the 'truncation' in t...

FWIW, there's never been a mention of parameter sets yet, just "three new parameters and annotations".

Hmmm... if your source is a number and your target is a number, why are you doing a TO_CHAR() in the source sql?

It's not documented though Ray does note your options there.

:!: OK you two, first off stop quoting everything every time you post. There's a perfectly lovely Reply to topic you should be using instead, which will also save me from having to clean up the mess all those unnecessary quotes leave. And for goodness sake, don't bury your reply in the middle of a h...

Well, crap... I thought I remembered that it supported a format mask but it doesn't, silly function. There are Server functions that will do that if you are willing to use a BASIC Transformer. Or still use the DecimalToString function with 'fix_zero' argument and you'll get values like this: 0000000...

Yes it does and yes it is. :wink: You should be able to read it as a string and use the StringToTimestamp() function to match the "default format" of a date/time in your database, talk to your DBA if you don't know exactly what that is. Or take the value as you have it now and use a TO_DAT...

A dataset does not need to be read with the same number of 'nodes' that it was written with.