what does this warining mean??

nag0143 · Post by **nag0143** » Sat Jul 17, 2004 4:56 pm

Hi,

What does this warning message suggest ??

Program "JOB.1623522433.DT.1334767455.TRANS1": Line 186, Nonnumeric data when numeric required. Zero used.

There are only three integer/decimal fields on my target, and i have checked those and even after looking at line 186 in the date i don't see any problem... where might have been the problem to generate this warning??

Thanks
Nag.

ray.wurlod · Post by **ray.wurlod** » Sat Jul 17, 2004 5:06 pm

Line 186 does not refer to the data. It refers to the subroutine called JOB.1623522433.DT.1334767455.TRANS1 (which you can view in the RT_BPnnn directory, where nnn is the job number reported in the error message).

This message usually arises from attempting to perform an arithmetic function or operation on data that turn out to be non-numeric. It could be, for example, incrementing a stage variable where the previous value of the stage variable is non-numeric.

Inspecting the subroutine will help you to diagnose this more accurately; you can figure out what the subroutine is trying to do at or near line number 186.

nag0143 · Post by **nag0143** » Sat Jul 17, 2004 6:16 pm

Thanks Ray,

Yeah there are some values in the column 17 like 234-
and in the xfm i am doing int(column17)/100 where actually it is expecting a number like 234 instead of 234-

Thanks
Nag

datastage · Post by **datastage** » Sat Jul 17, 2004 9:35 pm

You're lucky you had only 3 columns to look at. I've had that error before with like 50 columns that were candidates. It always seemed like DS should be able to return which column the error occurs on. It seems like Ray sorted you out, but if you have many columns and aren't sure what data might be the problem you have to break the transform up and have 1 job perform half the columns and another perform the other half. Find which one there error exists in and break that one in half... that will shorten the list quickly if you don't mind creating some temp jobs...

ray.wurlod · Post by **ray.wurlod** » Sun Jul 18, 2004 5:40 am

In this case, explore the Iconv() function in on-line help, and find out how to handle signed numbers with trailing sign. Use this function to convert the data to "conventional" format. Hint: it's part of the "MD" specification; find the MD topic in on-line help (or in the DataStage BASIC manual).

chinek · Post by **chinek** » Sun Jul 18, 2004 5:34 pm

This can also be caused when in the source data you might have CR/LF in your file at the end of the number. You can't see it but using the "od" command check your file to make sure that they are no trailing CR/LF in your numeric fields.

Nick