On Fri Mar 14, 2025 at 11:20 PM CET, Tamir Duberstein wrote: > On Fri, Mar 14, 2025 at 6:00 PM Miguel Ojeda > <miguel.ojeda.sandonis@xxxxxxxxx> wrote: >> >> On Fri, Mar 14, 2025 at 9:18 PM Benno Lossin <benno.lossin@xxxxxxxxx> wrote: >> > >> > I don't know when we'll be bumping the minimum version. IIRC 1.85.0 is >> > going to be in debian trixie, so eventually we could bump it to that, >> > but I'm not sure what the time frame will be for that. >> > >> > Maybe we can salvage this effort by gating both the lint and the >> > unstable features on the versions where it works? @Miguel, what's your >> > opinion? >> > >> > We could even make it simple, requiring 1.84 and not bothering with the >> > older versions. >> >> Regarding Debian Trixie: unknown, since my understanding is that it >> does not have a release date yet, but apparently mid May is the Hard >> Freeze and then it may take e.g. a month or two to the release. >> >> And when it releases, we may want to wait a while before bumping it, >> depending on how much time has passed since Rust 1.85.0 and depending >> on whether we managed to get e.g. Ubuntu LTSs to provide a versioned >> package etc. Yeah that's what I thought, thanks for confirming. >> If something simple works, then let's just go for that -- we do not >> care too much about older versions for linting purposes, since people >> should be testing with the latest stable too anyway. > > It's not going to be simple because `rust_common_flags` is defined > before the config is read, which means I'll have to sprinkle > conditional logic in even more places to enable the lints. > > The most minimal version of this patch would drop all the build system > changes and just have conditionally compiled polyfills for the strict > provenance APIs. Are folks OK with that? So you'd not enable the lint, but fix all occurrences? I think we should still have the lint (if it's too cumbersome, then let's only enable it in the kernel crate). --- Cheers, Benno
