DSXchange

So, can you have 2774389000000 in a Decimal(15,4) field?

Decimal(15,4) has a total of 15 significant digits. Four of these are to the right of the decimal point, which means you can have a maximum of 11 to the left of the decimal point.

Yes. Just use regular constraint expressions in the Transformer.

You do sometimes get the section mark in text pasted from Word documents. I prefer one of the Control Set 0 characters, such as Ctrl+A (which is Char(1) in DataStage) or Ctrl+Y (which is Char(25) in DataStage).

That's just how the interaction with Oracle with NLS enabled works. Depending on version specify four times or three times the actual number of characters in the metadata. Or use a specific string_from_ustring specification to effect the transfer.

Are you sure you've configured sparse lookup correctly? This signal is associated with memory. Could there be a NULL being returned but the receiving column is defined a Nullable=No?

How wide are your rows? There is a configurable maximum number of bytes per row - from memory it's 8192. There's also a configurable maximum number of columns per row - from memory that one is 400. Exceeding either of these might cause a heap allocation error.

TimestampFromSecondsSince() function with an offset of -86400 should also work.

You might try sorting the source so that the non-empty values appear first.

Otherwise run the file through a script (or a stage) that appends a dummy 21st column that is always populated.

The ODBC stage must have multiple inputs. Once that has been done a check box should become enabled allowing you to specify grouped transaction (if I recall correctly).

This is possibly a memory error. Can you provided sorted data to the Aggregator stage? If not, you can tune the size of the memory table used by the Aggregator stage.

Can you view the sequence log from the Director client?

The product (which is called Information Analyzer, incidentally) imposes no limits. Your operating system may. Also be aware that the Information Analyzer database needs to be sized appropriately - recommendation is four to five times the volume of data to be profiled.

My guess is that your email system doesn't allow attachments as large as 16 MiB. Most don't.

Each output link from the Transformer has three columns. The first is mapped directly from the ORDER_ID column. The second is derived - differently on each link - as a constant, one of the four remaining column headings, perhaps with case adjusted (GROUND --> "Ground"). The third is mapped...