DSXchange

Just the collector. If all you're trying to do is concatenate exact metadata matching files, just use a Unix cat statement. That's really fast and easy. Using a link collector requires both input links have the exact same metadata, so after your sequential stages you'll want a transformer to adjust ...

One possible solution: Put f1 into a hash file, stream f2 and reference f1 hash file, put constraint that compares all relevant columns and output the new or differences. Put f2 into a hash file, stream f1 and reference f2 hash file, put constraint that compares to see if found and output the new. S...

Link ordering is a critical issue. You will have to go thru every job and make sure your link ordering is correct. The reject link has to be the last link in the order. You will also have to make sure that all functions do not manipulate passed arguments. This change requires you to edit each functi...

We need to know, did the error happen on the primary input stream OCI stage or the reference stream OCI stage. I'm guessing the reference. If it is the reference, you're probably having an illegal datatype value come thru that doesn't match the datatype of the reference lookup. Just a guess without ...

You're talking about changing the radio button on the stage properties from reflecting account to external directory, and then filling in the external directory value with a job parameter. By my calculations, you have to figure out where this information is stored inside the dsx file and adjust acco...

mmmm corn tabs

OCONV(link.column, "MD20") will shift the decimal 0 places and round to 2 decimals. MD24 means shift 4 places and round to 2. MD00 means no shift and round up to no decimals.

Only thru opening each job in Designer.

Check your DS project's file system to make sure you're not out of space. If nothing runs, that's usually the first issue. Next, verify that userids/permissions haven't changed on the project/jobs.

Since the order of processing is basically read a row from the input, do all references, then write all outputs, all references must exist prior to any output. When a job starts, if the optimizer sees that some active stages (the second set) need to wait on passive references to be populated by othe...

You can run commands on remote machines using the remsh, rsh, or rlogin commands. It works for running commands on remote Windoze or Unix machines. The issue is permissions. It may sound like a gimmick, but you can always put triggers on a table that do the things you need done. Then, insert the app...

Then I suggest you execute a command line and run the SQL script or PL/SQL script that way. You have infinitely more options for error handling and complicated scripts.

The ODBC stage supports stored procedures, what's the big deal?

Search the forum for using SETFILE. However, be aware that this leaves a pointer in the project for that hash file. Sometime in the future you may relocate the hash file, and will need to update that pointer again.

Your hashed file stages have a value in the account name (really the project name) field in the stage properties. This means your hash files will always reside in that project. If you had left this field blank, the hash files would default to the current project in which the job is running. I suspec...