Revert "lib/trivial: fix 'error: cannot decode virtual path '/nix/sto…

[Artturin]

…re/virtual0000000000000000000000005-source''"

This reverts commit b67ee6e.

#202244

error: a string that refers to a store path cannot be appended to a path, at /etc/nixos/nix/nixos-unstable/lib/sources.nix:193:30

appears to happen when there's a nixpkgs git submodule

So one of the things that is different for a git submodule is that the .git folder isn't a folder, it's a textfile that contains (in my case) this:

$ cat nix/nixos-unstable/.git
gitdir: ../../.git/modules/nixpkgs

Description of changes

Things done

• Built on platform(s)
  • x86_64-linux
  • aarch64-linux
  • x86_64-darwin
  • aarch64-darwin
• For non-Linux: Is sandbox = true set in nix.conf? (See Nix manual)
• Tested, as applicable:
  • NixOS test(s) (look inside nixos/tests)
  • and/or package tests
  • or, for functions and "core" functionality, tests in lib/tests or pkgs/test
  • made sure NixOS tests are linked to the relevant packages
• Tested compilation of all packages that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage
• Tested basic functionality of all binary files (usually in ./result/bin/)
• 22.11 Release Notes (or backporting 22.05 Release notes)
  • (Package updates) Added a release notes entry if the change is major or breaking
  • (Module updates) Added a release notes entry if the change is significant
  • (Module addition) Added a release notes entry if adding a new NixOS module
  • (Release notes changes) Ran nixos/doc/manual/md-to-db.sh to update generated release notes
• Fits CONTRIBUTING.md.

[amaxine]

My system builds again, thanks.

[edolstra]

The use of toString <path> is undesirable for performance reasons (it causes an unneeded copy of Nixpkgs to the Nix store) and will cause problems once NixOS/nix#6530 is merged. I'm not sure whether the use of nixpkgs as a submodule should count as a blocker - is that a use case that we actually support?

[Artturin]

The use of toString <path> is undesirable for performance reasons (it causes an unneeded copy of Nixpkgs to the Nix store) and will cause problems once NixOS/nix#6530 is merged.

yep can't use a path nixpkgs input in a flake without getting error: cannot decode virtual path '/nix/store/virtual0000000000000000000000005-source' after reverting this commit
inputs.nixpkgs.url = "path:/home/artturin/nixgits/my-nixpkgs";

I'm not sure whether the use of nixpkgs as a submodule should count as a blocker -
is that a use case that we actually support?

it's common with niv i think
this might happen with non-nixpkgs submodules too

i opted to revert it for now because NixOS/nix#6530 is a nix PR while some peoples working systems can't rebuild now

[amaxine]

@edolstra I wouldn't advocate for blocking such a change on the basis of my or anyone else's use, but when I started using nix this was (at least in my eyes and based on friends' recommendations) common enough that I opted to follow the idea. If this behaviour is unintentional, I think it still warrants a deprecation notice and a planned removal - e.g. by 23.05 - rather than just dropping it overnight because it wasn't meant to be supported in the first place.

I don't follow nix development closely enough to have known that this is an intentional change before now, and I don't recall seeing any mention of this in release notes either - so to me, this comes out of nowhere and breaks my systems with no prior warning.

[etu]

Yay, now it can unblock my updates of the unstable channel 🙂

And I agree with @maxeaubrey, maybe this could be based on state version or something?

Otherwise, if this change should be pushed for all of a sudden. It should at least be documented in the release notes that this is an intended change.

[etu]

I'm not sure whether the use of nixpkgs as a submodule should count as a blocker - is that a use case that we actually support?

Well, it's a use case that has been working fine for many years. So there's certainly people that uses it. Which means that it should at least be deprecated in some form of graceful way.

[github-actions]

Successfully created backport PR #202375 for release-22.11.

[talyz]

I'm not sure whether the use of nixpkgs as a submodule should count as a blocker - is that a use case that we actually support?

The code that broke was there specifically to support fetching the nixpkgs git revision from a submodule, so yes, it has certainly been a supported use case.

The use of toString <path> is undesirable for performance reasons (it causes an unneeded copy of Nixpkgs to the Nix store)

In this case, it seems toString was used to get the real path to .git and avoid copying the whole repo to the nix store.

[talyz]

Here is an alternative solution that could work with NixOS/nix#6530. It does however make a useless copy of the .git directory to the store.

[Artturin]

Here is an alternative solution that could work with NixOS/nix#6530. It does however make a useless copy of the .git directory to the store.

you forgot to link/paste the code

[talyz]

Oops! Here's the link: talyz@1e659be