Hi there,
I just started testing Aurora and either I don’t get the concept or I don’t know how to handle it.
Maybe you can help me.
I get that a read-only system enhances the reproducibility and thus lowers the risk for upgrade related errors.
With layering you can roll back to a previous state - that is nice too.
But the almighty internet states that rpm-ostree install
should only be taken as the last option if not otherwise possible.
So, I should install my stuff in local containers.
But whats the benefit of cluttering you users-space containers over cluttering your system over the time?
Or in other words, when I do the same mess in user-space?
Additionally, how would you recommend doing this? As it comes with all kinds of other quirks (ampersands no working properly from the “outside”, PATHs overlay from the “inside”, …).
So far I have just one container with the necessary basics (ripgrep, pdf reader, etc.) I always want to use - i.e. everything not present as a flatpack.
Do I configure the shell to always automatically enter the container?
And is there something lighter than fedora-toolbox
? I don’t need the whole system twice.
Questions over questions … Maybe there is a best-practice manual somewhere …
I think my ideal setup would be a system described as a config file (or config.d folder), where installing means adding something to the config so that I always have a coherent state.
Thanks a lot!