[samestep]

Resolves #4.

[samestep]

@me-and could you try this out and let me know if it works for your use case? I stopped short of adding a fully general evaluation config string argument because there were a couple tricky bits there, although I believe we could definitely add it later. And then the other deviation here is that, since we do still construct such a string internally, we just hash that string instead of normalizing it in some way (which would have had its own issues).

[me-and]

I'm having a play now; it might take a few days before I can do some comprehensive testing.

[me-and]

Although I've just done a relatively minimal test and, checking the build logs, it appears to be doing exactly what I'm after!

npb · 847c87de24b991b6dacf9eb8fe8c2cd0b89d4c96 → 847c87de24b991b6dacf9eb8fe8c2cd0b89d4c96*

Expand this for a reproducible command.

npb --base 847c87de24b991b6dacf9eb8fe8c2cd0b89d4c96 --head 847c87de24b991b6dacf9eb8fe8c2cd0b89d4c96 --patch /dev/stdin --overlay 'final: prev: { gitMinimal = prev.gitMinimal.override { doInstallCheck = false; }; git = prev.git.override { doInstallCheck = false; }; gitSVN = prev.git.override { doInstallCheck = false; }; gitFull = prev.gitFull.override { doInstallCheck = false; }; }' --no-tests -p gitMinimal -s x86_64-linux <<'PATCH'
diff --git a/pkgs/by-name/gi/git/package.nix b/pkgs/by-name/gi/git/package.nix
index 9c656cbfa85a..ff171f19aecc 100644
--- a/pkgs/by-name/gi/git/package.nix
+++ b/pkgs/by-name/gi/git/package.nix
@@ -56,7 +56,7 @@
   cargo,
   rustc,
   nix-update-script,
-  withBreakingChanges ? false,
+  withBreakingChanges ? true,
 }:

 assert osxkeychainSupport -> stdenv.hostPlatform.isDarwin;
PATCH

Expand this for a legend of all symbols below.

• ✅ = successfully built
• ❌ = failed to build
• 🚫 = dependency failed to build
• ❔ = couldn't try to build
• ⏩ = failed to evaluate
• ➖ = doesn't exist

x86_64-linux

✅ → ✅ · 1 package

• gitMinimal

[me-and]

...and having spotted the error in my previous command (gitSVN = prev.git.override... rather than gitSVN = prev.gitSVN.override...), rebuilding with that corrected had npb repeat the "resolve" stage but then immediately produce the same results since the underlying derivations weren't changed by that fix.

Looking in the ~/.cache/nix-npb/ directory, I can see the second run added a second cache directory.

So that's all exactly what I'd expect in this scenario 🎉

[samestep]

Amazing! :) Hopefully I'll have some time in the next couple days to get this PR ready to merge and then make a v1.2.0 release.