Update libchallenge bypass ristretto ffi#106
Conversation
it depended on now-deleted nix source file make it use niv sources instead also drop the NIXPKGS_REV which has been unused for a while
|
This is in good shape for review except that it should be changed to point at a master revision of python-challenge-bypass-ristretto once there is one that has the currently targeted branch merged (see LeastAuthority/python-challenge-bypass-ristretto#42). |
tomprince
left a comment
tomprince
left a comment
There was a problem hiding this comment.
This looks fine to land after the upstream PR has landed.
I do have a suggestion about nixpkgs pins, but that should be handled in a separate PR.
| # time of this comment. We can bump it to a newer version when that | ||
| # makes sense. Meanwhile, the platform won't shift around beneath us | ||
| # unexpectedly. | ||
| NIXPKGS_REV: "92609f3d9bc3acffbdbe54fa1c591a885612aa73" |
There was a problem hiding this comment.
I guess the only place this was being used was for the fetchFromGitHub call in nix/challenge-bypass-ristretto-repo.nix so is now unneeded.
| sources = import ./sources.nix; | ||
| in | ||
| import "${src}/default-challenge-bypass-ristretto-ffi.nix" { } | ||
| { callPackage |
There was a problem hiding this comment.
I guess this means that we use whatever pinned version of nixpkgs that we get from haskell.nix for compiling the rust library. I wonder if we should instead be using our own pin, which we can update on our own schedule (in practice, we'd pass in the version from PrivateStorageio for deployment, I think).
I think for the haskell packages, it makes sense to use the haskell.nix pins, since that we we can use their binaries, and it matches what they test. But they update the pin less often than we do, so it would make sense to use our pin for things not related to haskell.
However, this is out-of-scope for this PR.
There was a problem hiding this comment.
Ah yea, that sounds like a reasonable future change.
2 participants
Fixes ever-green libchallenge_bypass_ristretto_ffi link errors. Probably for good this time, maybe!