DSXchange

FIELD(In.StringColumn,"=",2) will give you what is after the '=' and before the next '=' (if present). Is that what you were looking for? If not, please also specify if this is a Server or a PX transform, as Server has additional string manipulation functions which are not present in PX. ...

I had a copy of V7 running on that, never tried it with V8.

Ray, as usual, is correct. I did a quick test/check but only on hashed files, then checked ODBC and didn't find the attribute and misremembered that it doesn't work, but now I've looked at the transform stage and see a tick box for "Reference link with multi row result set".

One method is the unix tr command, i.e.

As I stated in the previous post, server jobs output 1 row for each input row regardless of how many matches. It is a lookup stage and not a join stage.

DataStage doesn't do this kind of conversion, but you could replace the commas in the CSV with tabs and get a formatted output if the column lengths are identical or within one tab of each other. Another is to make HTML output, which most mail readers can interpret: x.csv: Col1,Col2,Col3 1,2,3 HTML ...

In this case the source or reference stage types are not relevant, server doesn't support this type of functionality. You can do this in PX jobs with the lookup or join stages, though.

I haven't tried it, but since it is not a supported platform the odds of getting any help from IBM approach 0.

A timestamp does not contain time zone information, so you need additional data in some form to (a) find out what time zone a given timestamp value is in, and (b) define 'global time' for your application and add/subtract the requisite number of hours from your timestamp. What database (some support...

You marked this as a server job but placed it into the PX forum. From the question, I'll assume that this is indeed a server job. Since hashed files cannot have duplicate keys your reference cannot have 3 records with the same key column.

How do you determine the relative ages of the files? Is it part of the file name, or an OS attribute (date-time-created or date-time-modified or the like)?

ds_dwh - I think you need to research into what fact and dimension tables are and what DataStage actually does with data before progressing further. The DataStage product has no idea of what fact or dimension tables are; the product deals with data at a lower and more granular stage, moving and tran...

Again, the heap message is a warning and not an error. Write a test job with just the CFF stage and output to a PEEK stage. As you've done already, get it to read and display one column successfully using "view data". Once that is working, add columns until you get an error and then post t...

I would do an explicit conversion so that both datatypes are identical in type and you will need to ensure that nullability is identical as well, perhaps by adding a NullToValue() call.

If your DataStage partitioning is identical to your database partitioning then, when loading to the database, each process only needs to load to one partition and is therefore much faster and more efficient than otherwise. This is not necessarily DataStage but general database and, in this case, Ora...