Since I am on the roll now with again testing different setups and ways I got thinking the best optimal setup for atomic development.
Let’s say we use password manager that managea SSH,CLI, browser extension with GIT and Commit
Is it just easy and simple way to layer all those to base or should you go for flatpaks and permissions?
For me I now days try to use systems as they are meant to and layer only VPN on atomics rest is experimental and testing what works on what setup.
I am almost on there password managers improving finally to get flatpak support and talks and integrations on board and where I now hit small issue is actually get containers to talk to flatpaks and IDE’s I haven’t found yet what makes me happy now
I’d like to use we storm, but flatpak is so outdated and crashing all the time, VScode works, but I want to avoid it if I can
Atomic desktops are not “meant to not be used with RPMs”.
Packages are better as RPMs or as Flatpaks. That si the criterium.
Actually I would be more careful using RPMs on traditional Fedora than on Atomic. the BTRFS is unfinished so no snapshots, not talking about automatic.
And not ostree so not resettable, transparent overlays etc.
The only issue is speed, and I run Kinoite with like 10 packages layered on an Intel i3 Chromebook, no issue at all.
If you dont need to permanently install temporary packages, like in development environments, layering is just okay.
For stuff like compiling stuff or building software I would use a container. Same with software that had a ton of dependencies like QGis in my case. But would be more efficient to layer that too.
About permissions, this is an ecosystem question. Example: VPNs on Android are brilliant. the stuff like always-on and even “killswitch” are built into the android system. Apps can have a VPN permission, which means the systems sends all traffic to them. They do whatever they want, but cannot manipulate your system.
On Linux, there is no such thing. The barebones networkmanager implementation doesnt have a killswitch and doesnt allow plugging in flatpak apps to do the VPN stuff.
Well I have VPN with threat protection enabled, kill switch enabled, firewall, IPv6 and ipv4, virtual locations and many other settings
I use flatpaks mostly as android only layers are VPN and NVIDIA
I know containers and I use them mostly getting the almost done idea where I hot the wall that missing connection link is not there…
I am okey for now that not fully integrated password manager cli and ssh stuff, but what I am not okey is to find the correct workflows to setup the development.
I setup multiple containers one for rust one for 3D etc or just use all in one. Main thing I want is to use editor/ide on host and connect that to containers since much easier when running multiple containers to connect. The issues comes all the hacks and wrappers to need to create this shouldn’t be actually so complicated thing to get permissions and overrides to actually get stuff working
I am thinking to test nvim setup if I can make it valuable solution, but again that might need me to layer it on host still
You still have the issue then that you count on one container for all your stuff and that is not the optimal solution since I don’t want all in one container
i have idea since there is VI already as base so that to upgrade as default to Nvim would be nice that way we can just go and setup Nvim and toolboxes and start coding this seems to be pretty straight forward and valid aproach
since there you have Nvim as default base already and just setup configs and done you can have multiple containers and switch how you want…
for now i cant see any IDE or Editor to have these options
I would be also just happy if there is default editor setup already that has all correct rights and permissions to detect toolbox containers Kate on KDE Kinoite or that upgraded superpower NeoVim
What works for me is VSCode flatpak + Homebrew for CLI tools. Alternatively, you can use Toolbox/distrobox for CLI tools, but IMO homebrew is slightly easier to use and then reset if necessary.
I would really try to not layer CLI tools onto your OS if at all possible.
there seems to be something wrong on this aproach as in general the idea and tools are there, but no actual documentation how to achieve what they want…
atomics are here already and then created toolbox to help get better apps and development workflows and then created container workflows, but how it is supposed to be use as eyes of creators and those who created them.
is this again the part where tools are here, but not actually fully working and yous till need to count VScode to use toolbox containers
i am seeing here the pattern lots of ideas and good ideas and those are done, but actually where those leads is a mystery since there is no fully working scheme on ideas