Jenkins Declarative Pipeline without Jenkins file in the repository

Question

I am trying to upgrade my current regression infrastructure to use pipeline plugin and I realize there are two methods: scripted pipeline and declarative pipeline. Going through multiple articles, I realize that declarative pipeline is more future proof and more powerful, hence I am inclined to use this. But there seems to be following restrictions which I don't want to have in my setup:

1. The jenkinsfile needs to be in the repository. I don't want to keep my jenkinsfile in the code repository.

2. Since the jenkinsfile needs to be in SCM. Does it mean I cannot test any modification in the file until I check that in to the repository.

Any details on the above will be very helpful.

Answer 1

Declarative pipelines are compiled to scripted ones, so those will definitely not go away. But declarative ones are a bit easier to handle, so all fine for you.

You don't have to check a Jenkinsfile into VCS. You can also set up a job of type Pipeline and define it there. But this has the usual disadvantages like no history etc.

When using multi-branch pipelines, i.e., every branch containing a Jenkinsfile generating an own job, you just push your changed pipeline to a new branch and execute it. Once it's done, you merge it.

This approach certainly increases feedback cycles a bit, but it just applies the same principles as when writing your software. For experimentation, just set up a Pipeline type job and play around. Afterwards, commit it to a branch, test it, review it, merge it.