| 16 May 2024 |
 @jade_:matrix.org | (we do not think this is a good thing) | 02:31:57 |
| @jade_:matrix.org | you know what else was broken for ages and the docs lied about? overriding nested flake inputs. we fixed that bug tho :P | 02:32:16 |
 Qyriad | In reply to @jade_:matrix.org
(you can guess what i am about to tell you)
(the nix team wontfix'd this) yeah we saw. anyone wanna pick it? | 02:34:36 |
 FireFly | ok so I guess it's pick your poison:
nix-instantiate --eval --json --argstr path "$1" nixos-option.nix | jq -r
nix eval -f nixos-option.nix --apply "f: f { path = \"$1\"; }"
| 02:34:37 |
| FireFly | ohwell | 02:35:03 |
 nevivurn | There is no widely used / official lix-install-action yet, right? | 02:36:09 |
| @jade_:matrix.org | In reply to @nevivurn:nevi.dev
There is no widely used / official lix-install-action yet, right? we really should just fork the detsys one | 02:37:23 |
| @jade_:matrix.org | In reply to @firefly:raccoon.college
ok so I guess it's pick your poison:
nix-instantiate --eval --json --argstr path "$1" nixos-option.nix | jq -r
nix eval -f nixos-option.nix --apply "f: f { path = \"$1\"; }"
you would have to do more explicit things to make flakes work with it | 02:37:47 |
| @jade_:matrix.org | In reply to @jade_:matrix.org
we really should just fork the detsys one this isn't a super high priority but if you want to help with it, i can get a fork put up? | 02:38:26 |
| @jade_:matrix.org | i think it would just need some docs rewriting and of course change the defaults | 02:38:43 |
| FireFly | I don't really speak flakes so I think I'd be happy with just getting rid of the c++ blob I think (re nixos-option) | 02:40:17 |
| @jade_:matrix.org | yeah i would say it would be a strict improvement even to not support flakes | 02:40:33 |
| FireFly | I mean it should be easy enough to improve in tiny-shellscript+nix form :p | 02:40:46 |
| @jade_:matrix.org | and we could try harder later. yeah. | 02:40:52 |
| @jade_:matrix.org | omg ok . wow. the nix installer action machinery from detsys is a terrifying bit of engineering | 03:06:32 |
| @jade_:matrix.org | i forked it but omg https://git.lix.systems/lix-project/lix-install-action | 03:06:40 |
| @jade_:matrix.org | we would have to fork "detsys-ts", "determinate-nix-shim" docker package, and possibly some other bits | 03:10:07 |
 julia | why is there vendored tarballs generated by docker | 03:10:16 |
| julia | why | 03:10:18 |
 puck | In reply to @jade_:matrix.org
omg ok . wow. the nix installer action machinery from detsys is a terrifying bit of engineering don't you like 1000 lines of typescript in one file | 03:10:22 |
| @jade_:matrix.org | actually we could just delete the docker stuff because it's for gh enterprise server only | 03:10:49 |
| @jade_:matrix.org | also why is it using docker instead of like. shoving it in a ns jail itself or .. why is it even doing this? | 03:11:20 |
| puck | also, doesn't forgejo have very similar actions | 03:11:24 |
| @jade_:matrix.org | yeah probably | 03:11:33 |
| @jade_:matrix.org | we do want such a thing | 03:11:41 |
| puck | In reply to @jade_:matrix.org
also why is it using docker instead of like. shoving it in a ns jail itself or .. why is it even doing this? it's not always nested docker, also | 03:11:44 |
| puck | sometimes you get the host's docker socket | 03:11:55 |
| @jade_:matrix.org | In reply to @puck:puck.moe
sometimes you get the host's docker socket right, but, uhhhh, host root, | 03:12:10 |
| @jade_:matrix.org | like i don't think we should allow that configuration at all | 03:12:19 |
| @jade_:matrix.org | but idk | 03:12:22 |
