Sync source on DataStage job

benny.lbs · Post by **benny.lbs** » Fri Mar 10, 2006 2:13 am

We have parallel development on DataStage job and would like to Sync source on DataStage job. However, it is found that it is not easy job.

Case 1:

I have a Job in project A, then export it as DSX (A.dsx) and import to project B, compile it in project B, and export it as DSX (B.dsx). The source should be no changes, but when 'diff' these two DSX file, there are still changes except 'project name, Time'

Case 2:

We have parallel development on Core job, they have different release schedule, so Sync source is required. However, it is not quite easy as there are lots of option within a job and hard to be trace all. Besides, if it is a Transformer, the code is generated by system, also hard to be compared.

Can anyone share your experience on Sync source ?

ArndW · Post by **ArndW** » Fri Mar 10, 2006 5:33 am

Benny,

have you thought about using DataStage version control to help synchronize your work? It would simplify development and make sure that only one version of a job is considered "live".

ray.wurlod · Post by **ray.wurlod** » Fri Mar 10, 2006 6:25 am

There's some good news for you in the Hawk release, where they have a "difference" tool that allows you to compare two DataStage objects side by side. Hyperlinks to where the actual differences occur are part of what you get.

chulett · Post by **chulett** » Fri Mar 10, 2006 7:26 am

Case 1:

If you are asking for confirmation - yes, that's how it would work. There will be some changes as you mentioned, like project name and create/mod times, and others that should be of no consequence. What differences are you seeing that concern you?

Case 2:

I wouldn't even consider doing 'parallel development' on a single job. There's no mechanism to 'merge' the changes back together other than what sits between the chair and the keyboard. It's not like text based coding where an intelligent source management system can help straighten out the mess for you.

You may be able to get away with having two copies of the job, one for each development effort, but it would require very disciplined developers. Or can the job be broken down into more modular segments, something that supports the parallel development efforts by isolating functionality that may change?

benny.lbs · Post by **benny.lbs** » Fri Mar 10, 2006 9:51 am

Thanks all of you first.

ArndW,

I am using MKS for version control, however, as chulett said that

"It's not like text based coding where an intelligent source management system can help straighten out the mess for you"

Ray,

Where can I get more information about your mentioned tools ?

ArndW · Post by **ArndW** » Fri Mar 10, 2006 10:05 am

A typical source control system with it's check-in and check-out would only work on .DSX or .XML flat files extracted from or loaded into the projects. I've seen this done but it was quite unwieldy and, IMHO, unmanageable for the longer.

ray.wurlod · Post by **ray.wurlod** » Fri Mar 10, 2006 11:24 am

The Hawk release is still in beta. There are email addresses within IBM where they might be prepared to give you more information. There were handouts at the Las Vegas conference. But I'm on the road, and not carrying them with me, so can't help. Try asking your account representative to get them for you.
I think it will be late this year at the earliest that the next release will become generally available.