!UUqahLbShAYkkrXmKs:matrix.org

DevOS

33 Members
Seeking help and geeking out together on https://github.com/divnix/devos & https://github.com/divnix/digga10 Servers

Load older messages


SenderMessageTime
2 Sep 2021
@blaggacao:matrix.orgDavid Arnold (blaggacao) grahamc (he/him): would the operating theory also work for sd iso? (havn't had time to look at it in greater detail) 15:03:56
@blaggacao:matrix.orgDavid Arnold (blaggacao) * grahamc (he/him): would the operating theory also work for sd and other iso? (havn't had time to look at it in greater detail) 15:04:05
@grahamc:nixos.org@grahamc:nixos.orgif you're ever making a "real" filesystem like ext4 or whatever: no15:05:09
@grahamc:nixos.org@grahamc:nixos.orgif you're saying well we don't need that, let's just do a ramfs: it could work, but you'd need to figure out putting down a bootloader15:05:31
@blaggacao:matrix.orgDavid Arnold (blaggacao)I was associating with the time it takes to build the squashfs..15:05:47
@grahamc:nixos.org@grahamc:nixos.orgagain, a squashfs is too much of a real FS 15:06:20
@grahamc:nixos.org@grahamc:nixos.orgthere are other approaches you could take15:06:40
@grahamc:nixos.org@grahamc:nixos.orgbut they have downsides15:06:45
@grahamc:nixos.org@grahamc:nixos.orgthis is the second time I've tried to make a fast ipxe server for nixos15:06:54
@grahamc:nixos.org@grahamc:nixos.orgthe first time I took a squashfs-based approach15:07:12
@blaggacao:matrix.orgDavid Arnold (blaggacao)manveru pointed to the link today... He's sitting in quite the same boat.15:07:31
@grahamc:nixos.org@grahamc:nixos.orgevery store path got its own squashfs, each squashfs got its own cpio, and then I'd loop over all the squashes and mount them:15:07:33
@grahamc:nixos.org@grahamc:nixos.orghttps://github.com/grahamc/netboot.nix/blob/master/quickly.nix#L49-L7415:07:34
@grahamc:nixos.org@grahamc:nixos.orgthis repo uses recursive nix to share those cpios / squashes between builds to reduce duplicate work15:07:50
@grahamc:nixos.org@grahamc:nixos.orgbut recursive nix was not a very good experience, and still had a major issue of UX of copying around big initrd's from build machines to the machine hosting the imafges15:08:24
@grahamc:nixos.org@grahamc:nixos.orgby keeping everything Nix does small, the nix-build and nix-copy-closure time is very small15:09:15
@b12f:pub.solarb12f

After updating to devos' master branch, we're running into the following issue:

❯ nix flake update
warning: Git tree '/home/ben/Workspace/os' is dirty
error: input 'bud/beautysh/poetry2nix/flake-utils' follows a non-existent input 'beautysh/flake-utils'
(use '--show-trace' to show detailed location information)
15:48:43
@b12f:pub.solarb12fAnyone know where this might be coming from? We've tried going back and forth between different branches, deleting lockfiles, etc.15:49:11
@b12f:pub.solarb12f

This doesn't work

      bud.inputs.beautysh.inputs.poetry2nix.inputs.flake-utils.follows = "digga/flake-utils-plus/flake-utils";

But I guess something like this would be the way to fix this, no? Interestingly, our flake.nix inputs are basically the same as for devos master

15:58:27
@timdeh:matrix.org@timdeh:matrix.orgIt should work in theory at least15:59:37
@b12f:pub.solarb12f

results in the following:

error: cannot find flake 'flake:beautysh' in the flake registries)
16:00:18
@ultranix:matrix.org@ultranix:matrix.org what do i do about error: input 'bud/beautysh/poetry2nix/flake-utils' follows a non-existent input 'beautysh/flake-utils' while trying to update flakes 22:38:45
@blaggacao:matrix.orgDavid Arnold (blaggacao) nix develop github.com:divnix/digga --command nix flake update 22:41:55
@blaggacao:matrix.orgDavid Arnold (blaggacao)may that could work...22:42:02
@blaggacao:matrix.orgDavid Arnold (blaggacao)the problem is the local unpatched nix version...22:42:19
@ultranix:matrix.org@ultranix:matrix.orgyeah i figured22:43:10
@blaggacao:matrix.orgDavid Arnold (blaggacao)since that patch was merged, pain will be temporary.22:43:47
@b12f:pub.solarb12fI think for us what did it was running `nix flake lock --update-input <name>` at least the build has now been running for a while22:55:40
@b12f:pub.solarb12f`<name>` with the name of an input, one after the other down the list22:56:44
@tomberek:matrix.orgtomberek This came up in another channel as well. The UI for “nix flake update NAME” means that the NAME flake’s inputs are fetched. And this ends up being a registry lookup. And the most common failure case is “nixpkgs”. Because it resolves and looks as if it updated. But not your input in the current directory, but nixpkgs in your registry. 23:13:22

Show newer messages


Back to Room ListRoom Version: 6