DSXchange

When we write to Oracle using the OCI stage, with both the bulk and upsert modes any rejected writes (i.e. when a PK constraint has been violated) will generate a well-hidden informational message but no warning. I just noticed this issue and am trying to find a way to easily generate warnings. I se...

Phil,

thanks for the update; I'll have to remember that "gotcha" with the merge stage should I have odd errors when I use it.

Doesn't the DSJobReport use @FM as a separator? You can do CONVERT(@FM,'~',YourString) to check

Benny, send a request for an enhancement in to Ascential via your support contract. This has been a request for a while now, Ascential is fully aware that this would be a great additional to the functionality. But, as always, the squeaky wheel will get the grease and with the limited Engineering res...

The Precision attribute will influence your display, what is that set to?

Mark, (as has been stated before, there is no such thing as an urgent request in this forum) Your mainframe files are EBCDIC, so the first step is to convert them to ASCII. Usually when you get files from the host you will use FTP, which can be configured to automatically do this conversion; or you ...

Phil, Since the quotes are balanced you are getting another cause. When DS parses the input columns it uses the column delimiters (which don't occur within quoted strings) as parsing tokens Plus the end-of-line and end-of-file delimiters. An odd error as you have written the files yourself. Can you ...

Which stage are you using to which database?

Reddy, the function would be CONVERT("'|","''",In.BigColumn) . Please note the string delimiter in the first parameter is a double quote and the second paramater uses a single quote. This function takes all occurrences of ' and | and replaces them with " and " respectively in the string In.BigColumn

Since the sequential file stage only allows one delimiter you can't do it in just one step. I would write a job that defines this sequential file as having one column and no field delimiters at all. Run this stream through a transform that does a REPLACE of all pipe and single-quote characters into ...

blewip, in your transform stage equation, please remember that your string of 1234.5678 is also being converted into a float during the transform (an implicit conversion, that is most likely to single precision) - and that a chances are quite low that one float will exactly equal another. Before rai...

Blewip, I would say that 5 digits of precision is completely unacceptable. Something is going seriously wrong. Single-precision floating point is (I think) usually 32 bits for the mantissa and this should allow for much more accuracy. Please take a sampling of a couple (perhaps 5) different real num...

I think that NLS is not recommended in your case; but that you will need to make some decisions & do some coding in order to handle some of the functionality the NLS would give you. DataStage allows you to specify your quote character in sequential files. If you choose a quote character that als...

A bank financial application never uses floating point representation to store financial information. The books will never balance. I remember doing a project (years ago) for the Dresdner bank international currency trading division and they were using quad precision and still coming up with daily t...

Blewip, the way that floating point numbers are stored are with a exponent and mantissa (plus a sign). A floating point number is usually internally represented as your mantissa x (e to the exponent). It is highly unlikely that any given number will be able to exactly equal any fixed-length mantissa...