DSXchange

Your questions appear unconnected. Are they for any interview ? yes..... Well, at least that's an honest answer. Problem is we're not here to answer interview style questions and make someone appear to know something they in fact do not. We're more than happy to help solve legitimate problems, howe...

Why do you think you 'have to' know this?

As noted, via the Sort stage and you can click the Help button in the stage for usage notes. Biggest thing is to understand the 'Sort Key' syntax.

Ah... Windows. UNIX wouldn't care. You'll need to add the path separator in the container it seems - I assume you are doing "#par1##filename#" there, yes?

It doesn't work as advertised? No clue, never actually used it in all these years, prefer Oracle sequences for concurrent work, which I prefer to avoid. What 7.x release are you on, btw?

Short answer is 'no'. Longer answer involves including the record number as a data element and doing an SSELECT or 'sorted select' to bring them back out in the input order. Requires an account-based hashed file, from what I recall, however.

It's not getting changed. What is your key in the hashed file? How unique is it? You are aware of the 'destructive overwrite', last-one-in-wins way they work, yes?

So, they fixed something that never should have been working. What makes you think your change will 'make all other jobs not run well'? Have your tried it? What happened?

We ended up switching from an xmlbeans "svalidate" solution as it wasn't catching all of the nit-picky things that Google was in the xml we were generating and sending to them. We would pronounce it valid and they would pronounce it crap. :( Ended up having our Java Boys build something ar...

Mark the output field of your sum as Nullable: Yes.

As noted, a script would be fairly simple. Do a 'for' loop thru the files, echo the filename to the output file and then cat the file itself. Use '>>' redirection in each case to append rather than overwrite.

AFAIK, you would need to logon to the UNIX server, cd $DSHOME and view .odbc.ini to check the IP or HostName for your entry. Everything else would be set in the stage. That or attempt to import ODBC metadata from the Manager using the connection.

So... you need to generate it like that from the start or can you generate them separately and then combine them afterwards? The latter is just 'concatenation' in a loop and something you could script. The former... would have to ponder that.

ps. Next question: are you 100% certain the table the job is populating is the exact same one you checked in the analyser tool? Easy to check the wrong database or schema or whatnot.