Revamp nur-package-template #109

Pandapip1 · 2025-07-07T18:25:52Z

Fixes #88

Fixes #89

Fixes #96

TODO:

Pandapip1 · 2025-07-07T21:26:35Z

I'm half tempted to merge this now, but I'll wait until a week has passed. It makes nur-packages-template a lot more opinionated, which I think is probably actually a good thing.

ToyVo · 2025-07-12T15:16:48Z

README.md

-
-## README template
+3. Configure CI: Change your NUR repo name and optionally add a cachix name in
+   [CMakeLists.txt](./CMakeLists.txt).


Link to the correct file that is imported by ci and the templates

Suggested change

[CMakeLists.txt](./CMakeLists.txt).

[config.yml](./config.yml).

ToyVo · 2025-07-12T15:26:41Z

maintainers/ci.nix

  isReserved = n: n == "lib" || n == "overlays" || n == "modules";
  isDerivation = p: isAttrs p && p ? type && p.type == "derivation";
-  isBuildable = p: let
-    licenseFromMeta = p.meta.license or [];
-    licenseList = if builtins.isList licenseFromMeta then licenseFromMeta else [licenseFromMeta];
-  in !(p.meta.broken or false) && builtins.all (license: license.free or true) licenseList;
+  isBuildable =
+    p:
+    let
+      licenseFromMeta = p.meta.license or [ ];
+      licenseList = if builtins.isList licenseFromMeta then licenseFromMeta else [ licenseFromMeta ];
+    in
+    !(p.meta.broken or false) && builtins.all (license: license.free or true) licenseList;
  isCacheable = p: !(p.preferLocalBuild or false);
  shouldRecurseForDerivations = p: isAttrs p && p.recurseForDerivations or false;

-  nameValuePair = n: v: { name = n; value = v; };
+  nameValuePair = n: v: {
+    name = n;
+    value = v;
+  };

  concatMap = builtins.concatMap or (f: xs: concatLists (map f xs));

-  flattenPkgs = s:
+  flattenPkgs =
+    s:
    let
-      f = p:
-        if shouldRecurseForDerivations p then flattenPkgs p
-        else if isDerivation p then [ p ]
-        else [ ];
+      f =
+        p:
+        if shouldRecurseForDerivations p then
+          flattenPkgs p
+        else if isDerivation p then
+          [ p ]
+        else
+          [ ];
    in
    concatMap f (attrValues s);

  outputsOf = p: map (o: p.${o}) p.outputs;


would it make sense to move these functions to lib/default.nix with the other internally used functions and add a with nurAttrs.lib;?

I considered this. The reason I decided against it was because I didn't want people removing the functions from lib/default.nix (which is meant to be user-editable) and breaking their config. I could still be convinced otherwise but I'll keep it like this for now.

ToyVo · 2025-07-12T15:30:32Z

maintainers/ci.nix

+    )
+  );

 in


add the with nurAttrs.lib; here

ToyVo · 2025-07-12T15:32:39Z

overlay.nix

  isReserved = n: n == "lib" || n == "overlays" || n == "modules";
-  nameValuePair = n: v: { name = n; value = v; };
+  nameValuePair = n: v: {
+    name = n;
+    value = v;
+  };


if we moved the other functions, then these functions will already be available using with nurAttrs.lib;, or inherit

ToyVo · 2025-07-12T15:37:09Z

flake.nix

+            formatter = pkgs.treefmt.withConfig {
+              runtimeInputs = with pkgs; [
+                nixfmt-rfc-style
+                black
+                yamlfmt
+                markdownlint-cli2
+              ];
+
+              settings = {
+                on-unmatched = "info";
+                tree-root-file = "flake.nix";
+
+                formatter = {
+                  nixfmt = {
+                    command = "nixfmt";
+                    includes = [ "*.nix" ];
+                  };
+                  black = {
+                    command = "black";
+                    includes = [ "*.py" ];
+                  };
+                  yamlfmt = {
+                    command = "yamlfmt";
+                    includes = [ "*.yml" ];
+                  };
+                  markdownlint = {
+                    command = "markdownlint-cli2";
+                    options = [ "--fix" ];
+                    includes = [ "*.md" ];
+                  };
+                };
+              };
+            };


suggestion: use https://github.com/numtide/treefmt-nix
could look like

treefmt = { programs = { black.enable = true; mdformat.enable = true; nixfmt.enable = true; yamlfmt.enable = true; } }

I've intentionally tried to reduce the number of flake inputs here; I was until recently using numtide's treefmt module in NUR but it got removed in a recent PR.

fgaz · 2025-07-12T18:48:33Z

maintainers/scripts/render_templates.py

+        return yaml.safe_load(f)
+
+
+def render_all_templates(src_dir=".", config_path="config.yml"):


you might want to use https://staticjinja.readthedocs.io, it's in nixpkgs

Interesting, I'll check it out!

ToyVo · 2025-07-16T23:32:33Z

maintainers/default.nix

+{
+  stdenvNoCC,
+  makeWrapper,
+  python3,
+}:
+
+{
+  render_templates = stdenvNoCC.mkDerivation {
+    name = "render_templates";
+
+    src = self;
+
+    nativeBuildInputs = [ makeWrapper ];
+
+    installPhase = ''
+      runHook preInstall
+
+      mkdir -p $out/bin
+      makeWrapper ${python3.interpreter} $out/bin/render_templates \
+      --add-flags "${./scripts/render_templates.py}"
+
+      runHook postInstall
+    '';
+  };
+}


I tested this following the suggestion by @fgaz https://github.com/nix-community/nur-packages-template/pull/109/files#r2202868549

Suggested change

{

stdenvNoCC,

makeWrapper,

python3,

}:

{

render_templates = stdenvNoCC.mkDerivation {

name = "render_templates";

src = self;

nativeBuildInputs = [ makeWrapper ];

installPhase = ''

runHook preInstall

mkdir -p $out/bin

makeWrapper ${python3.interpreter} $out/bin/render_templates \

--add-flags "${./scripts/render_templates.py}"

runHook postInstall

'';

};

}

{

jinja2-cli,

writeShellScriptBin,

}:

{

render_templates = writeShellScriptBin "render_templates" ''

find . -name "*.j2" -print0 | while read -d $'\0' file; do

${jinja2-cli}/bin/jinja2 "$file" config.yml --format=yml > "''${file%.j2}"

done

'';

}

anttiharju · 2025-12-30T17:49:26Z

.github/workflows/build.yml

    strategy:
      matrix:
-        # Set this to notify the global nur package registry that changes are
-        # available.
-        #
-        # The repo name as used in
-        # https://github.com/nix-community/NUR/blob/master/repos.json
-        nurRepo:
-          - <YOUR_REPO_NAME>
-        # Set this to cache your build results in cachix for faster builds
-        # in CI and for everyone who uses your cache.
-        #
-        # Format: Your cachix cache host name without the ".cachix.org" suffix.
-        # Example: mycache (for mycache.cachix.org)
-        #
-        # For this to work, you also need to set the CACHIX_SIGNING_KEY or
-        # CACHIX_AUTH_TOKEN secret in your repository secrets settings in
-        # Github found at
-        # https://github.com/<your_githubname>/nur-packages/settings/secrets
-        cachixName:
-          - <YOUR_CACHIX_NAME>
        nixPath:
          - nixpkgs=https://github.com/NixOS/nixpkgs/archive/refs/heads/nixpkgs-unstable.tar.gz
          - nixpkgs=https://github.com/NixOS/nixpkgs/archive/refs/heads/nixos-unstable.tar.gz
          - nixpkgs=https://github.com/NixOS/nixpkgs/archive/refs/heads/nixos-25.05.tar.gz
    runs-on: ubuntu-latest


To help nix-darwin users, I wish the runs-on value could be defined with the nixPath.

See what I've done in my nur-packages repository: https://github.com/anttiharju/nur-packages/blob/09429286cd108e174d0065a74404cd8d78a3c747/.github/workflows/build.yml#L13-L24

strategy: matrix: include: - nixPath: nixpkgs=https://github.com/NixOS/nixpkgs/archive/refs/heads/nixpkgs-unstable.tar.gz runner: ubuntu-latest - nixPath: nixpkgs=https://github.com/NixOS/nixpkgs/archive/refs/heads/nixos-unstable.tar.gz runner: ubuntu-latest - nixPath: nixpkgs=https://github.com/NixOS/nixpkgs/archive/refs/heads/nixos-25.11.tar.gz runner: ubuntu-latest - nixPath: nixpkgs=https://github.com/NixOS/nixpkgs/archive/refs/heads/nixpkgs-25.11-darwin.tar.gz runner: macos-26 runs-on: ${{ matrix.runner }}

I think it would be fair to leave macos-latest entry commented out by default because macOS runners are fairly expensive, but making it easy to toggle on would be much appreciated by nix-darwin users I think 🙏

Happy to submit a separate PR though, this one appears to have been inactive since Jul 17

It must have fell off my TODO. I'll see about getting this done.

Pandapip1 force-pushed the flake-parts-revamp branch 2 times, most recently from 60771fe to bfe4802 Compare July 7, 2025 20:00

Pandapip1 added 5 commits July 7, 2025 16:17

Revamp nur-package-template

dc16774

Add new example-package

8bd96ab

Update README to remove travis section

e7fb570

And add markdownlint to nix fmt

da93776

Create config.yml as a central place to set the initial config

0715588

Pandapip1 force-pushed the flake-parts-revamp branch from bfe4802 to 0715588 Compare July 7, 2025 20:21

Pandapip1 added 3 commits July 7, 2025 17:18

Begin reseperation of code

94bd76b

Remove getFlake.nix (now unused)

0be2c53

WTF, why does *that* fix eval

158a708

Pandapip1 marked this pull request as ready for review July 7, 2025 21:25

Pandapip1 added the community-feedback-wanted label Jul 7, 2025

This was referenced Jul 7, 2025

Use by-name #88

Open

Copyright notice modification in repositories based on this template #96

Open

ToyVo reviewed Jul 12, 2025

View reviewed changes

fgaz reviewed Jul 12, 2025

View reviewed changes

ToyVo reviewed Jul 16, 2025

View reviewed changes

anttiharju reviewed Dec 30, 2025

View reviewed changes

	[CMakeLists.txt](./CMakeLists.txt).
	[config.yml](./config.yml).

+                  )
+                );
               in

		return yaml.safe_load(f)


		def render_all_templates(src_dir=".", config_path="config.yml"):

Uh oh!

Revamp nur-package-template #109

Are you sure you want to change the base?

Revamp nur-package-template #109

Uh oh!

Conversation

Pandapip1 commented Jul 7, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Pandapip1 commented Jul 7, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

Pandapip1 commented Jul 7, 2025 •

edited

Loading