DSXchange

I know I've done that before as well. Can you click (left or right) in the menu bar space to see if you get an option? Worst case you might just reinstall the designer or check and remove the registry entries. I don't have DS on this machine to cross-check, sorry.

That is odd - you have 2 columns separated by a comma in your source but need to append a 2nd "," in order to read this file? Have you specified fixed width - as indicated by the error message?

I have one here that I could loan you

Seriously, which part of the Designer screen is gone? The actual menu text bar, the graphics or something else?

Find out which ASCII character they map to, and use CHAR(nnn) to see if the error goes away. Which one of the 3 is causing the issue?

I didn't know what code 141 meant, but I used the search button to find it.

DSJE_LOGTOFAILED -141 Failed to log to UV account

What is your dsjob syntax? Are you returning a unix code or a DataStage one? Are you certain you are actually running the job?

"did not work" doesn't help much in analyzing the problem. What happened? Did the machine crash? Was the output just 0000-00-00? Are you 100% certain that this date column is generated the error messages? Do you know which value of Date is causing this? You can narrow it down to the row - using IsVa...

This is a case for using Stage Variables. You don't need to TRIM() each input column 3 times. Do it once in a Stage Variable and use that value. A valid date would have a LEN of 10 and have 2 '-' characters in addition. What about using the IsValid() function - have you tried that?

The DataStage client programs can only be installed on the supported Windows versions - they cannot be installed on UNIX.

CHAR is going to be same length, 1 byte per CHAR, on all systems unless you are working with multibyte NLS representations, in which case one CHAR can be multiple bytes. Actually, are you sure of you data types? The strings you listed in your original post are very reminiscent of binary data (i.e. C...

sendmkpk/devidotcom - both of you are correct , I was wrong. Even though I had written DCOUNT I was thinking in terms of the COUNT function. COUNT("yyyy-mm-dd","-") is 2 and DCOUNT("yyyy-mm-dd","-") is 3. The "#2" part means "<>2". It makes sense to use the COUNT function in this place instead of DC...

The error message is clear. Without RCP you need to declare your source and target columns explicitly; in this case you haven't declared "Col1". When RCP is turned on the job assumes that a column exists when you use it, so this error would disappear. Since you didn't state what problems you are get...

Yes, it is possible to pass a column name as a parameter. I don't understand the link portion of your question. You can create a modify stage to create a new column and use the parameterized column name in that modify stage.

sendmkpk - yes, you are wrong. How many "-" characters does a correctly formatted yyyy-mm-dd string have?

If you remove the column with the StringToDate() does the error go away or remain?

Can you explain how a TimeStamp value can fit into CHAR(10) with an example where there is no truncation of data occurring during the conversion? Your search should be more into why the development environment is not producing this error.