You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I personally believe that we do not need this attribute in buildRustPackage. It lets us slightly simplify code and documentation, but mainly it's because, despite efforts, I always experience OCD-like compulsion to make it consistent.
I have read NixOS/rfcs#131 and see that reasonable minds differ, so I offer this issue as a discussion for Rust infra and package maintainers focused only on our Rust ecosystem.
I'm in favor of doing this and willing to put up some PRs for it (even if it only is until we find a way to move away from these hashes and their abuse of FODs altogether), so I'm especially interested in hearing objections. I'll seed the main concerns against doing this that I see expressed in that RFC:
To support copying hash values in non-SRI formats from external sources, old versions of nix, or automation scripts.
To maintain backwards compatibility.
to which I'll add a third:
Not worth the time (e.g. reviewing, rebuilding, potentially fixing breakages).
/cc @NixOS/rust
The text was updated successfully, but these errors were encountered:
I'm ok with requiring usage of cargoHash instead of cargoSha256 in newly added Rust packages, but I don't think actively changing existing packages is worth the time.
I personally believe that we do not need this attribute in
buildRustPackage
. It lets us slightly simplify code and documentation, but mainly it's because, despite efforts, I always experience OCD-like compulsion to make it consistent.I have read NixOS/rfcs#131 and see that reasonable minds differ, so I offer this issue as a discussion for Rust infra and package maintainers focused only on our Rust ecosystem.
I'm in favor of doing this and willing to put up some PRs for it (even if it only is until we find a way to move away from these hashes and their abuse of FODs altogether), so I'm especially interested in hearing objections. I'll seed the main concerns against doing this that I see expressed in that RFC:
to which I'll add a third:
/cc @NixOS/rust
The text was updated successfully, but these errors were encountered: