DSXchange

No, he meant literally what he said - a single new link from the Transformer marked as a Reject link and set to abort after 1 would have accomplished the same thing. While what you did may work, it is complete overkill.

How about a few more details about what the database is and what you're trying to accomplish here?

That's because in order to do the 'all or nothing' part of the rollback for each 'transaction group' it forces the Transaction Size to 1.. It's the slowest way to put data in tables but also the only way to accomplish this. Seems like your only opportunity for improvement comes with the 8.5 release....

Not so much a starter as an ender IMHO. The one I typically use is much... simpler.

No worries!

Meaning they should compare just fine.

What issue are you seeing? What makes you think that isn't happening properly? Can you provide some specific examples?

That external format should compare just fine. Can you provide your exact syntax, input values and output?

Thank you for posting your resolution.

That would be via a lookup and in the Server product that means hashed files. For example, load up a hashed file with the appropriate columns from TC_BANK_HOLIDAY, looks like that would be HOLIDAY as the Key and (at a minimum) HOLIDAY_CODE as a data column. Then in your transformer reference link as...

Please continue this conversation in your original thread, there's no need to open anything new.

You use a single target stage and multiple links from the transformer, then enable 'Transaction Grouping'. Master first via proper link ordering.

As noted, SQL in routines is not for the faint of heart and especially not one for someone new to the tool. Nothing here, however, sounds like a routine of any kind is required. These just sound like a series of cascading lookups to me with some if-then-else logic to decide which one to use.

So... which is correct, Server or Parallel?

An exact search for "Non-pipe (or non-virtual data set) with .v in its name" returned 28 other posts, did none of them help?

This sounds like a 'century pivot' issue but that shouldn't be an issue with 4 digit years, only 2 digits years where it has to determine the century. :? Just for grins, does anything change if you tweak your code like so: ereplace(Oconv(Iconv(ClmHdr.BIRTH_DT, "D-DMY[2,A3,4]"), "D-YMD...