delete the direct root manipulation of the nix store #18

New issue

Open

opened 2024-03-11 17:44:38 +00:00 by jade · 4 comments

jade commented

2024-03-11 17:44:38 +00:00

Owner

would fix: NixOS/nix#10139

if everything goes through the daemon it would increase reliability. we can't necessarily eliminate the need for this for e.g. nixos-enter though. sigh. spin a background daemon? idk.

would fix: https://git.lix.systems/NixOS/nix/issues/10139 if everything goes through the daemon it would increase reliability. we can't necessarily eliminate the need for this for e.g. nixos-enter though. sigh. spin a background daemon? idk.

jade added the

stability

label

2024-03-11 20:01:09 +00:00

qyriad commented

2024-03-11 20:12:38 +00:00

Owner

We don't need to fully delete the capability, just make it never the default (which it never should have been).

jade referenced this issue

2024-03-18 16:28:25 +00:00

[Nix#10139] nix lock can modify lock files in the nix store and corrupt it #47

ma27 self-assigned this

2024-06-24 10:06:12 +00:00

ma27 commented

2024-06-24 10:06:25 +00:00

Member

Self-assigning, will give this a shot :)

lix-bot commented

2024-09-26 20:49:25 +00:00

Member

This issue was mentioned on Gerrit on the following CLs:

comment in cl/1963 ("Fix passing custom CA files into the builtin:fetchurl sandbox")
commit message in cl/3569 ("cli: fork+exec subdaemons, don't just fork")

This issue was mentioned on Gerrit on the following CLs: * comment in [cl/1963](https://gerrit.lix.systems/c/lix/+/1963) ("Fix passing custom CA files into the builtin:fetchurl sandbox") * commit message in [cl/3569](https://gerrit.lix.systems/c/lix/+/3569) ("cli: fork+exec subdaemons, don't just fork")

lix-project referenced this issue from a commit

2025-01-29 18:11:12 +00:00

Merge pull request #18 from nix-community/meta-arg

ma27 commented

2025-03-27 18:58:18 +00:00

Member

Personally, I agree with this approach. And I don't think it makes sense to force us to a review cycle for an inferior solution before.

Hence, unassigning myself.

(Also, sorry that it took me so long to do something about it. Yes, I do feel kinda bad about it! :( )

So, I actually hacked a draft together on Oceansprint (it's actually kinda dangerous to change the behavior, I managed to turn the testsuite into a giant forkbomb with this). @raito mentioned that it's planned to get rid of single user mode in the first place and use on-demand daemons instead. Personally, I agree with this approach. And I don't think it makes sense to force us to a review cycle for an inferior solution before. Hence, unassigning myself. (Also, sorry that it took me so long to do something about it. Yes, I do feel kinda bad about it! :( )

ma27 removed their assignment

2025-03-27 18:58:26 +00:00