How to maintain your own fork of spark?

[petRUShka]

Couple of years ago I've cloned spark and have started to have my own changes in repository (because my set of software and settings couldn't be 100% same as somebodies else). From time to time I've cherry-pick`d some commits from upstream. But now I have ~2 years history of commits in upstream I want to incorporate to my fork of spark (and then somehow apply changes with ansible).

So it is more like question and not an issue. What is the best practice?

How to:

• have simultaneously my own fork with my own changes some roles are commented out and etc
• pull upstream changes
• have changes (from commits) applied to my installation

[pigmonkey]

The idea is that you would clone the repository, track changes from the master branch here (wether you do so in the master branch on your clone, or put them in some other branch), and make your personal changes in a different branch. Then you just merge changes from my upstream master branch to your personal branch.

GitHub has documentation on this sort of thing:

https://docs.github.com/en/github/collaborating-with-pull-requests/working-with-forks/configuring-a-remote-for-a-fork
https://docs.github.com/en/github/collaborating-with-pull-requests/working-with-forks/syncing-a-fork#syncing-a-fork-from-the-command-line

Ansible is idempotent, so you should be able to just run the playbook again to apply any new additions to your system.

Removing old things is a manual process. If I decide to drop a software package from spark, for example, I don't write a task to uninstall it. I just remove the lines from the task that installed it. So you'd have to manually look through the commits for things like that and pacman --remove ... the package. The same goes for removing files from spark. For example, the recent Tor change in 144226d. When you look through the commit you'd see that spark is no longer pushing the old Tor systemd override file. To match that change on your system you would rm -r /etc/systemd/system/tor.service.d.

[petRUShka]

@pigmonkey do you see some automated way to find all packages not installed by spark? And Same for configuration files in /etc and other places like that?

[pigmonkey]

No.