make changes to a shared container

[koojo]

I need to make changes to a shared container. The compile option in the designer is not enabled for me to make the change. How does one make changes to a shared container.

Do we need to copy the code on a new server job and then modify it and make changes out there and save it as a shared container

[chulett]

A Shared Container is a code 'fragment' and cannot be compiled on its own. Make your changes, save them and then see how it works out when you next work with a job that uses it.

[sun rays]

I need to make changes to a shared container. The compile option in the designer is not enabled for me to make the change. How does one make changes to a shared container.

Do we need to copy the code on a new server job and then modify it and make changes out there and save it as a shared container

if you make changes to the shared container, then all the other jobs which uses this shared container need to be compiled again.

[kcbland]

Do we need to copy the code on a new server job and then modify it and make changes out there and save it as a shared container

Yes, and that's about the only way to even write one. You have to be able to test it before using it in a larger job.




I really hate shared containers.

[koojo]

Thank you

[ArndW]

Ken, I think that all of those who have worked with older versions of DataStage have this deep-seated and almost irrational dislike of containers. I shudder and cringe whenever I have to use them, even though they no longer bear the sign of the devil

[ray.wurlod]

If you do change a Shared Container you will need to recompile every job that uses it. Usage Analysis can reveal which jobs these are.

[kcbland]

almost irrational dislike of containers

I think I'm quite rational. I rationally hate the implementation of the concept, but I'm cool with the concept. DS Functions are great examples of reusability. DS Transforms are NOT well implemented. After 9 years of major releases of DS you'd think engineering would put a Test button on the Transform just so you can see if it works. Shared containers have to be developed as jobs, then the "executability" has to be destroyed turning it into a container. To develop or enhance a container, you've got to turn it back into a job and repeat the process. That seems like a poor implementation of a reusable concept.

[ArndW]

I agree that engineering (as well as the rest of the company) has treated shared containers badly for years. At least now they work are no longer black holes for coding work - it used to be that once stages passed the 'event horizon' of a shared container they were lost forever and could no longer be pulled out. I have seen some rather odd containers used - where the container contains stages that require each row to be read in before one could be pushed out (i.e. a Hashed file stage or an aggregation) and this can cause quite a bit of confusion when running the job.