CI: Run the new test in CI

This requires removing the "ping Google" step (dunno if we don't have
ping or can't talk to Google specifically), but that only existed to
help distinguish "the test VM lacks network" from "Nixathena is broken",
and hopefully we've resolved the latter, so it's not a disaster.

Author: dehnert

.github/workflows/test.yml

@@ -0,0 +1,30 @@
+name: "Run tests"
+on:
+  pull_request:
+  push:
+
+jobs:
+  tests:
+    strategy:
+      matrix:
+        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-24.05.tar.gz
+    runs-on: ubuntu-latest
+    steps:
+    - name: Checkout repository
+      uses: actions/checkout@v4
+    - name: Install nix
+      uses: cachix/install-nix-action@V28
+      with:
+        nix_path: "${{ matrix.nixPath }}"
+        extra_nix_config: |
+          experimental-features = nix-command flakes
+          access-tokens = github.com=${{ secrets.GITHUB_TOKEN }}
+    - name: Show nixpkgs version
+      run: nix-instantiate --eval -E '(import <nixpkgs> {}).lib.version'
+    - name: Run flake checks
+      run: nix flake check
+    - name: Run standard metapackage test
+      run: nix run '.#test.meta-standard'

test/meta-standard.nix

@@ -19,7 +19,6 @@ pkgs.nixosTest {
     # network-online.service is too long to wait
     # afsd.service is (seemingly) long enough to have network
     machine.wait_for_unit("afsd.service")
-    machine.succeed("ping -c1 ipv4.google.com")
     [status, out] = machine.execute("ping -c1 google.com")
     print(out)
     machine.succeed('ls /afs/athena.mit.edu/')