lianad: init at 6.0

[plebhash]

Description of changes

the liana package currently only supports the liana-gui executable, while it is desired to also support headless use-cases

this PR adds a withGui attribute, which is used to choose between:

• provide liana-gui executable if withGui is true
• provide lianad + liana-cli executables if withGui is false (plus a $out/etc/liana/config.toml based on lianad_config_example.toml)

cc @dunxen

Things done

• Built on platform(s)
  • x86_64-linux
  • aarch64-linux
  • x86_64-darwin
  • aarch64-darwin
• For non-Linux: Is sandboxing enabled in nix.conf? (See Nix manual)
  • sandbox = relaxed
  • sandbox = true
• 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/)
• 24.11 Release Notes (or backporting 23.11 and 24.05 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
• Fits CONTRIBUTING.md.

---

Add a 👍 reaction to pull requests you find important.

[dunxen]

Thanks for opening this!

Left some comments, and it seems that only lianad gets built without liana-cli when withGui is false.

[dunxen]

it seems that only lianad gets built without liana-cli when withGui is false

@dunxen are you sure? this is what I get here:

$ nix-build -A lianad
/nix/store/jjldm51jm0jss2lnac841v4725hbjcg1-lianad-5.0
$ tree result        
result
├── bin
│   ├── liana-cli
│   └── lianad
└── etc
    └── liana
        └── config.toml

4 directories, 3 files

You are correct. I mischecked the result.

[dunxen]

Oops. Didn't see the PR title (I cannot edit it). Should be:

lianad: init at 5.0

[dunxen]

LGTM. Thanks!

[Aleksanaa]

it looks a bit weird because you are effectively making two derivation (which are different except for version and source.) into one. This may not be a good idea.

Good idea:

1. Use callPackages to declare multiple packages.
2. Just build them separately, and add a comment to remind keeping version the same
3. Same to 2 but add assert to keep version the same. This may break r-ryantm, however

[dunxen]

it looks a bit weird because you are effectively making two derivation (which are different except for version and source.) into one. This may not be a good idea.

Good idea:

1. Use callPackages to declare multiple packages.

2. Just build them separately, and add a comment to remind keeping version the same

3. Same to 2 but add assert to keep version the same. This may break r-ryantm, however

I agree in the current state it is not as discoverable / clear in the source what's going on, but I'm not sure which of those options more closely aligns to other packages that are in the same situation (see for example lshw vs bitcoin and a few others). Seems there are a few different methods.

I'm unfortunately unaware of any documentation around this practice, but would love to keep consistent and still have this maintainable without breaking r-ryantm.

Either way, if this should be a separate package declaration, I'll help if we need to keep them in sync manually without an assert (i.e. Option 2)

[Aleksanaa]

see for example lshw vs bitcoin and a few others

That's withGui, just an option to decide whether to include gui or not, and is often passed as an argument to build tools. The withGui in this package is actually building another package. They have different sourceRoots and different dependencies. Calling it withGui is quite misleading.

Yes, there are more cursed examples, see catppuccin (although I do want to remove that stuff for a long time)

[dunxen]

That's withGui, just an option to decide whether to include gui or not, and is often passed as an argument to build tools. The withGui in this package is actually building another package. They have different sourceRoots and different dependencies. Calling it withGui is quite misleading.

Yes, there are more cursed examples, see catppuccin (although I do want to remove that stuff for a long time)

Right, got it. Thanks @Aleksanaa for clearing that up! So let's not add to the cursed list then :)

I'm inclined to go with Option 2 and have a separate package declaration in pkgs/by-name/li/lianad/package.nix. @plebhash, is this alright with you? You can add me and yourself as maintainers there. Then you can just drop the diff in all-packages.nix :)

We can investigate option 3 at some later stage.

[dunxen]

Just changing the approval status to make it clear we're splitting this package.

[Aleksanaa]

Suggested change

	cargoSha256 = "sha256-6r8G4sO/aDO5VyJbmbdFz3W2b8ZmulXJZtnz8IAwcaU=";
	cargoHash = "sha256-6r8G4sO/aDO5VyJbmbdFz3W2b8ZmulXJZtnz8IAwcaU=";

[Aleksanaa]

Suggested change

	description = "A Bitcoin wallet leveraging on-chain timelocks for safety and recovery";
	description = "Bitcoin wallet leveraging on-chain timelocks for safety and recovery";

[Aleksanaa]

Please delete this

[dunxen]

Need to run nixfmt.

[dunxen]

Needs rebase and should init at 6.0 now. I had to update liana as 5.0 had build errors with rust 1.80: #333059

[dunxen]

Thanks! Last few comments 🤞

[dunxen]

First commit message should be ｀lianad: init at 5.0｀