How to divide reusable tasks yet using a single Docker container?

[Fabien Lydoire]

Hello there !

I am new to Bitbucket pipelines coming from Jenkins.
I want to have a pipeline with the following tasks:

1. Setup Javascript dependencies
2. Lint code
3. Generate doc
4. Execute tests
5. Report tests results to PR

I implemented those tasks using multiple step .
All steps are using the same Docker container which is cached and restored for each step. This process of caching/restoring is time consuming.
Also, those steps are defined and then used as YAML anchors and references to avoid duplication: reshuffled steps depending on the branch (a feature branch, develop, main, …).

My question: Is there a way to get all the following:

• Simple tasks reusable (just like step and YAML reference)
• All tasks using the same Docker container without cache/restore
• Visual representation in the Pipeline layout of the individual tasks to quickly identify where the issue, if any, was raised.

Thanks for your support !

[Erez Maadani]

Hi @Fabien Lydoire 

If your commands are one-line, just have one step per branch type/group and run them together.

If the commands are more complex, my recommendation is to implement these tasks externally to the pipeline/steps, IE, use a build script: gradle, maven, sh file, etc'. This will simplify the commands, or even group of commands, to one-liners and you can use the first option. 

[Fabien Lydoire]

one step per branch (...) and run them together

Yes I can do that, but it does not match my third requirement:

• Visual representation in the Pipeline layout of the individual tasks to quickly identify where the issue, if any, was raised.

Having a single step means having a single node in the Pipeline graph. I wish I could split tasks within a single node.

[Erez Maadani]

Here is the visual representation you would have:

You can see each command has its own line so it's very easy to see which one failed.

Basically, each step in the pipeline is running on a different container. if you need to share data between the steps, you would need to use cache/artifacts mechanism or some other way to inject the data into the containers (git clone/ftp/download/etc'). 

[Fabien Lydoire]

Thanks @Erez Maadani I understand that there are no better options.

I implemented my Bitbucket pipeline according to your suggestion.

I saves run time, yet I'm still frustrated with the single step green bullet in the pipeline graphical view.... 🤷🏻‍♂️