DSXchange

You can import file metadata in DataStage. Have you already declared your .xls as a system ODBC data source? Once you do that, import the column metadata from the Designer (import -> table definitions -> odbc)

Yes, I prefer putting in an explicit sort stage before the join and explicitly sorting or stating "don't sort, already sorted" for each join key.

Quite a few people swear by "ultraedit" and comparing files is quite easy with that tool.

Have you tried "'2008-12-20 00:00:00','2009-03-20 00:03:00'"

VOC stands for "VOC"abulary file and it is the central file in a project that defines what keywords and files are available. Your error message meant that a job tried to create a file which already existed in the VOC.

There are stages in 8.0 which are wave aware; in addition I have a buildop which does wave processing which also runs at 8.0 and above, so I am not sure what information you received. Again, if you don't do wave processing then don't worry about it.

I uses bzip2 and gzip compression quite a bit and haven't noticed this before. What UNIX are you working on?

As this is a UNIX and AIX thread started by someone else, please start your own thread for a windows problem.

Often when DS is not brought down completely and restarted you will not restart the dsrpcd and connectivity won't work. Bring DS down correctly (use seach in the forum for detailed steps to check status) and bring it back up.

Use a left outer join between your time dimension and data.

Unless you need to process waves, you can demote or ignore this error message. I seem to recall having set a buildop's wave attributes once so that this warning is not displayed, but can't recall how.

hint - if you don't know what a wave in this context is, you can safely ignore the warning.

The pages starting at IBM Websphere DataStage have lots of information and downloadable documents.

There are a couple of approaches possible. One would be to have a time dimension where you can SELECT all records between two dates and thus use a lookup or join in your job. Another option would be to write a DS/BASIC function which returns a list of dates. Neither is exceptionally difficult to do.

make a job copy with just the CFF and writing to a sequential file. Does that correctly display the Polish characters? DON'T use "view data".l

try "COUNT DS_JOBOBJECTS"