ray.wurlod wrote:Nice thought, but you can't create F pointers from DataStage to UniData.
No, not from DataStage, from an account setup for DataStage to use. Since dictionaries for files contain A, S, F, T, and I correlatives/descriptors, you will not only see physical data but virtual as well. When DataStage imports metadata, it imports the dictionary. You often have a lot of junk columns, the worst being T correlatives since they are actually accessing other hash files.
It is paramount that you only pull back the physical data in a hash file, no virtual columns. In addition, dictionaries are often malformed, incomplete, or outright wrong. It behooves you to have clean dictionaries that accurately depict the data within. For that, you can setup an account on the Universe/Unidata side for DataStage to use. There, setup F-pointers to point to the data files, but use locally created dictionaries. You will access that account rather than the normal account. This, IMHO, is the best way to make sure you get data, as well as the most performant extracts.