ivn

NixOS so I can keep my config in git. I have a single nix config for all my machines (desktop, laptop and server) so I can share configuration between them. I use it to configure both my system and my user config, my dotfiles, with home-manager. Even my neovim config is in nix thanks to nixvim.

I don’t think I could go back now. It can be a bit of a pain from time to time and the learning curve is steep but it has so many advantages. Being able to rollback between config versions (called generations), having a consistent config between my machines, having it all in version control… The repo have so many packages and when there is a module it’s really easy to add a service. Writing new packages (derivations) and modules is also not that hard. It can be as simple as calling nix-init.

Had my main ssd fail on me a few month back and it was very simple to just replay the config and just get everything working as before. I only had to do the partitioning by hand (it can be done by nix but I’ve not gotten around to it yet). That’s why I only backup data and home partitions, not system partitions.

That’s not my claim.

Lol, “do your own research”.

This is not about me but about the advice you give to people. If you give advices you need to back them up, you can’t just answer “I disagree” to contradiction.

I’m not saying you said it was precise. I’m just saying that not every layer is important, or even good.

What if there are less people using that VPN IP than using that ISP IP, now your are giving more information with a VPN than without. How do you verify the claim that a VPN does not log? Spoiler, you can’t verify, it’s just trust. You have to chose who you trust enough to give them access to the website you visit.

You’re dodging the issue. Why would an IP that’s shared and does not give precise location be as important as precise GPS location?

No, those are not “layer of abstraction” and they do not have equal importances, claiming otherwise is counter-productive. Each data does not reveal the same quantity of information and their hierarchy depends on your threat model.

I don’t think that’s true for mobile IPs since they are shared.

And it’s very handy for this, I have the same config for all my devices (desktop, laptop and server). Enabling and disabling different modules depending on the host it’s deployed to.

That’s why they are asking for a self-hosted solution, not a Google one.

Snowflake is an entrypoint into the tor network, not an exit point. I’m not a lawyer but I don’t think there are any legal implications, or maybe in Russia or Iran. And the whole point is that its traffic is very hard to identify.

They block with DPI so DoH would not be enough.