DSXchange

I can't check here regarding the CFF stage, but the 3 numbers you posted in your thread are all negative, right? If the CFF won't handle it correctly (and I'm 95% certain it can, since this is basic COBOL functionality), then at least the conversion for negative numbers is a straightforward computat...

hargun wrote:...
END AS CURR_BALANCE
...

"CAST(END as float) AS 'CURR_BALANCE'"
and declare your variable as s float in DataStage.

If ODBC only support sfloat and not dfloat, then you need to CAST the column in your SQL.

CAST(CURR_BALANCE as float) AS 'CURR_BALANCE'

Actually, it looks like you need to use dfloat (double-precision floating point rather than single-precision). But either floating point representation is going to give a potential error, since numbers are represented as approximations (mantissa and exponent). It is always preferable to use decimal ...

Thanks Craig - I looked at the code again:

somu_june wrote:...
vRowcount = DSGetLinkInfo (JobName,StageName,LinkName,DSJ.LINKROWCOUNT)
...

should read:

Since you are using DSJ.ME as the job handle, a lot of the code is superfluous. I can't check right now, but I seem to recall that the link row count information isn't available until after the job has finished, but I'm not 100% certain - it would explain your routine's behaviour.

The datatype "smalldatetime" in SQL-Server would be a DataStage datatype of "timestamp". One cannot just subtract one from the other and expect that the program knows you want the answer as an integer number of months (instead of days, or decimal fractions of a year, or minutes, ...

At the moment I can't think of a way, but I am afraid I don't use the front-end query tools much; but perhaps someone more knowledgeable will step in here.

DSJ.ERRFATAL is a constant, the value is numeric. As a constant, it cannot be changed, thus the second parameter to these 2 user-written routines is an input-parameter. Without source code or more information, what is done with that value in the routines is a matter of conjecture.

Often the quickest way to do this is to export the jobs into a text file (either .dsx or .xml format) and then use your favorite text editor, preferably one with macro capability.

No, that is not possible.

Can you try to delete both jobs from the director or designer, then re-import just that job in both environments and compare? That would ensure that you get rid of any potential artefacts.

In your designer you can select "File -> Force Compile" from the menu.

Welcome to DSXChange, There is a hashed file called "DS_AUDIT" in each project with contains some information about the job creation and modification dates. This file is not documented, but can be read from a DataStage server job. Unfortunately the information about the actual user doing t...

rski - even without going into programming in c or c++, the first code shown only sets XXXXX.City for a non-null value. Something is missing there.

What if you were to explicitly do a "force-compile" from your designer? Is the generated code different or the same as before?

Unfortunately the pointer-to method won't work in this case, since DataStage uses it's own overloaded functions and definitions, i.e. the APT_Ustring class with methods to convert to other data types. I would look into changing the NLS mapping in the job itself so that DataStage won't try to do the ...