New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Hash mismatch on nixpkgsSrc #1648
Comments
This isn't really anything to do with haskell.nix, though? |
That depends. This hash is specified by the haskell.nix project in its |
I've been looking into this more. It doesn't reproduce on NixOS with Nix 2.3.16 nor Nix 2.11.0, aarch64-darwin with Nix 2.8.1 nor x86_64-darwin with Nix 2.10.3 on APFS. So it seems pretty specific to my setup. I've checked all the filenames in nixpkgs for case collisions or differences under NFC/NFD but this hasn't turned up anything. I'm pretty puzzled 😕 |
I think that nix might extract the files before hashing so that the source does not matter (tarballs and fetchgit for instance get the same hash if all the files match). |
Clever helped me troubleshoot this further. Turns out a single bit was flipped in a hash in a package description. So this is entirely on my hardware corrupting a file during fetching or unpacking or something. Sorry for the noise! |
See NixOS/nix#847 (comment) and NixOS/nix#2475 My guess is that a unicode character slipped into the nixpkgs source and picked a commit while it was still in there. Hopefully someone has removed it. Let's try bumping the bootstrap nixpkgs to our 22.05 pin. |
Making it match the current 22.05 pin in flake.lock. Was going to do this because of #1648, but might save some people a nixpkgs download so let's do it anyway.
* Bump nixpkgs pin used in default.nix Making it match the current 22.05 pin in flake.lock. Was going to do this because of #1648, but might save some people a nixpkgs download so let's do it anyway. * Fix test issue * Use flake.lock * Use flake.lock
@hamishmack, thanks for the response. That was my original suspicion but it turned out not to be true. As I said in my previous comment, this was entirely on my hardware. A single bit flipped in the fetched and unpacked source, changing a "p" to a "q" in a hash in a random package in I hope this didn't cause too much undue work on your part! ❤️ |
This is happening for me on x86_64-darwin. A common source of hash mismatches on Darwin is HFS+, which is case-insensitive by default. But this is just a tarball, so a single file and I don't believe that can make the difference.
The text was updated successfully, but these errors were encountered: