Merging Version of Jobs

anm1966 · Post by **anm1966** » Wed Jun 24, 2009 5:09 am

Is it possible to merge 2 different versions of a job? The reason for this is management want to have 2 different development streams that will start with the same initial version of a job. Each stream will produce a new version of a job. At a later date a decision will be made about what will be installed depending on the progress of each development.

Ie

Start with Job V1

Dev Stream A takes Job v1 and produces version Job v1.A
Dev Stream B takes Job v1 and produces version Job v1.B

Merge Job v1.A with Job v1.B to give Job v1.AB

This would allow them to make a decision on what 2 install

Ie Dev Stream A code
Or Dev Stream B code
Or The Merged code.

ArndW · Post by **ArndW** » Wed Jun 24, 2009 5:16 am

No, there is no facility in DataStage to "merge" two job versions. I wouldn't know where to begin doing this automatically, there are so many different combinations of changes possible that no software could know which branch to use for which stage of a job. You could always compare V1 and V2 of job and see the differences, but any reconciliation would need to be done manually (and would involve a high level knowledge of the job in question).

Sainath.Srinivasan · Post by **Sainath.Srinivasan** » Wed Jun 24, 2009 6:16 am

What do you expect as result of merging?

Do you expect
a.) two independent streams
b.) a single stream with setA followed by setB
c.) remove duplicate and redundant mapping followed by optimization?

For a and b is easy. For c, you can hire a consultant.

anm1966 · Post by **anm1966** » Wed Jun 24, 2009 6:41 am

Sainath.Srinivasan wrote:What do you expect as result of merging?

Do you expect
a.) two independent streams
b.) a single stream with setA followed by setB
c.) remove duplicate and redundant mapping followed by optimiz ...

Hi Si

The management want 1 job with all the fuctionality from Stream A combined with all the functionality of Stream B

Cheers
Al

chulett · Post by **chulett** » Wed Jun 24, 2009 6:43 am

The output wanted doesn't really matter, as Arnd noted the answer is still 'no' for any kind of automated solution. You'll need to do it manually.

Sainath.Srinivasan · Post by **Sainath.Srinivasan** » Wed Jun 24, 2009 7:33 am

You can look into building Shared Containers.

That way you can produce reusable components which can then be plug-and-play.

But note that the metadata in the links must be correct.

This will solve if you choose a or b options.

chulett · Post by **chulett** » Wed Jun 24, 2009 7:39 am

Sai, just want to make sure you're understanding the question. This isn't about data "streams" or data at all, this is about having two groups making changes to "the same job" - i.e. starting from a common point and creating two new versions from it - and then needing to merge these two changed jobs back into one job.

ray.wurlod · Post by **ray.wurlod** » Wed Jun 24, 2009 4:58 pm

WHY do they want this?

<rant>Resist stupid requirements!</rant>

anm1966 · Post by **anm1966** » Thu Jun 25, 2009 6:58 am

Thanks everyone for your comments. You were just confirming what I already knew

ray.wurlod wrote:WHY do they want this?

<rant>Resist stupid requirements!</rant>

Good Question Ray. Management...... No I dont want to start I will end up ranting on for hours.