Hash File useage

[mcolen]

Can a hash file be created and used in the same jobstream ie create hash file from oracle 9i and lookup in transform of teradata?

[ArndW]

mcolen,

yes, you can write to and read from a hash file in the same job, just make sure that you have buffering on both writes and reads turned off.

[mcolen]

Arnd

While I know you can read and write it in the same job I am seeing jobs for the first time creating a file and trying to use it at the same time and was wondering if this is correct.

[ArndW]

Yes, even if the file is being created in the same job it will work, provided the stage that creates the file is in the stream before the stage that reads from it.

[chulett]

And there's a little trick you can play with a Transformer as a source when your lookup comes first.

[neena]

And there's a little trick you can play with a Transformer as a source when your lookup comes first.

Can u please reveal the trick

Thanks
RR

[chulett]

You can start an input stream with a Transformer under certain circumstances and that's what you need to do here.

Add a Transformer and a link to the reference hash. A transformer can be used to generate data as long as you define a stage variable in it (don't need to use it, just define it) and then constrain the number of output rows or it will run forever. So...

Populate the link with the same column metadata that your hash needs. Doesn't matter what you put in the derivation fields but obviously you need to put something there. Set an @FALSE constraint on the link so no rows ever go down it. However, its presence will tell DataStage that you need to write to it before the reads happen and you can use that link to create the hashed file on the first run or to clear it at the beginning of a run.

Hope that makes sense.