DSXchange

It's a problem in installing the stage, or perhaps in finding its associated library. You may need to re-install the plug-in using the Package Installer, or to register the plug-in.

You can certainly, and safely, call Job1 twice in a row, as you suggest in your sequence. The second invocation does not start until the first invocation has finished, so it's never trying to run two instances. However, if Job1 aborts the first time it will not be in a runnable state the second time...

The most likely cause (he stated, leaping to all sorts of assumptions) is that one of your data sources has trailing space characters and the other does not. For a lookup to a hashed file to succeed the two values must match exactly as character strings. For example, 0004 will not match 4 (even thou...

I can see where Shree's confusion arose. I meant "online" as "immediately available". I grew up (and was educated) before the invention of the World Wide Web, even before ARPAnet. No terminals for us - it was all punched cards and batch runs, and spoolers and despoolers were real hardware that used ...

There is no single location. Such a document is promised in the next ("Hawk") release. Chapter 12 of the Orchestrate User Guide provides some interesting information about the format of Orchestrate error messages which may aid your diagnosic processes, particularly the section on which environment v...

Don't know. Reset is available. Have you tried it?

199901 is not a valid date in YMD format. There is no month 99. Threrefore it is returning the original data. The routine was written according to your original specification, which showed dates in YMD order. Adapt it to suit. You could do other comparisons for other formats in the same routine, but...

You don't need to throw away the original value. Apply an Iconv() function and test with Status() to see whether it was a valid date. If it was, use Oconv to convert that converted value into the desired format; otherwise pass through the original result. FUNCTION ConvertIfDate(Arg1) If UnAs...

Read all about it in on-line help (a good start point is the BASIC Tasks topic) or in the DataStage BASIC manual (basic.pdf).

It would be easy to do with a server job (or a BASIC Transfomer stage). The Iconv() function is largely agnostic to the format of a date, except for the order of the components (see Date Conversion with ICONV for more information).

There is a qualiification you can make to decimal types, which is "Allow All Zeroes". A packed decimal field completely filled with zeroes is normally illegal; this setting overrides that and allows an all zeroes packed decimal field to be understood to represent 0.0. This may help, although you say...

Ah, memories. The Axcel PACK was a chargeable component prior to 5.2 that did nothing more than enable multi-instance capability. From version 5.2 (after much grizzling in customer land about value for money) it was built in to the base product and not separately charged. You WILL need it to run mul...

This can happen if you import only the executable job from an export file (without importing the design-time components). Or it may mean, by whatever mechanism, the ROOT record for that job has been removed from the DS_JOBOBJECTS table. With expert help (for example Advanced Technical Support at IBM...

In short, what has changed? "Nothing" is not the correct answer. The change may have been wrought outside of DataStage. Do you have any information about what kind of exception was caught?

ProfileStage is about telling you what's there (profiling), not what to do with it. You go through a systematic series of analyses: column analysis to determine the type and nature of data in each individual column (for example domain of values, proportion of nulls, etc) using the entire table (a sa...