T-spec meeting 2025-03-20

--- title: T-spec meeting 2025-03-20 tags: ["T-spec", "meeting", "minutes"] date: 2025-03-20 discussion: https://rust-lang.zulipchat.com/#narrow/stream/399173-t-spec/topic/Meeting.202025-03-20 url: https://hackmd.io/ws8REEONQXCuoacMHzFFfA --- Attendees: - Joel Marcey - Pierre-Emmanuel Patry - Eric Huss - TC - Josh Triplett - Sid Askary - Jack Regrets: - Niko Matsakis Agenda: - Updates to the agenda? - FLS integration announcement update - Work Session - Continue PR review from last week ## Updates to Agenda Eric: Questions about a few other PRs ## FLS integration announcement update Making progress on the public announcement. Ferrous Systems is reviewing the draft and the goal is get to consensus early next week. Joel: We sent out the draft announcement to Ferrous Systems last week. They're reviewing, and hopefully that'll be ready next week. The idea is that we'll publish it on the Foundation blog and on its other social media entities last last week. Then we'll be working with the move of the repository over from Ferrous Systems to the project. Josh: Should we be planning for a blog post on the Project side? Joel: Maybe we should do that once the integration is complete? Josh: It seems preferable to just do a simultaneous announcement. From the Foundation side it makes sense to play up the partnership and similar. From the Project side, we can work on that text. TC: +1. Josh: Let's see a copy of that draft announcement. That'd be helpful for drafting the other. Joel: Will send that out. ## PR Review https://github.com/rust-lang/reference/pull/1703 -- modules: describe both module filename styles without giving a clear preference - ❓ Should we close? After some discussion, we decided to close this PR. TC to draft up wording in the PR and then close it. https://github.com/rust-lang/reference/pull/1707 -- Doc: Add the LoongArch stabilized target features - Updates required - ❓ Verify we are OK with stale English documentation? Eric to ping the LoongArch folks. ### Medium https://github.com/rust-lang/reference/pull/1761 -- Mention that “every address” ≠ “every pointer” Agreement that this seems like a strict and good improvement. Josh added small wording suggestion. https://github.com/rust-lang/reference/pull/1758 -- Modify mutable deref rule for &mut T Eric will tweak and merge https://github.com/rust-lang/reference/pull/1755 -- Add: orphan rule rationale Eric will move to note section (since it is informative) The problem is the ambiguity -- it becomes a breaking error to make any changes if it was allowed. Preserving the space for your own types and traits. Haskell also has orphan rule, with extension to bypass it. https://wiki.haskell.org/index.php?title=Orphan_instance https://github.com/rust-lang/reference/pull/1704 -- Correct repr(Rust) overlapping fields rule for enums and unions - Clarifications of repr(Rust) and unions https://github.com/rust-lang/reference/pull/1679 -- Explicitly specify syntax forms for each attribute https://github.com/rust-lang/reference/pull/1664 -- Add Values and Representation chapter - Probably waiting on ehuss https://github.com/rust-lang/reference/pull/1654 -- Spec: Improve Definition of Type Layout - Probably waiting on ehuss https://github.com/rust-lang/reference/pull/1635 -- Adjust identifiers in inline-assembly.md to be more consistent with other chapters https://github.com/rust-lang/reference/pull/1545 -- Spec abi chapter - Probably waiting on ehuss ### Unknown https://github.com/rust-lang/reference/pull/1752 -- Guarantee soundness of pointer-to-int transmutes - UCG, Ralf had concerns, who is responsible? Try to get more info and additional participation from Ralf. https://github.com/rust-lang/reference/pull/1703 -- modules: describe both module filename styles without giving a clear preference - This did not reach lang consensus? I am confused. https://github.com/rust-lang/reference/pull/1661 -- Define raw pointer transmute behavior - Lang-nominated - Not sure what's going on ### Waiting on author https://github.com/rust-lang/reference/pull/1707 -- Doc: Add the LoongArch stabilized target features - Updates required https://github.com/rust-lang/reference/pull/1701 -- sanitizers: Add documentation for the no_sanitize attribute - Updates required https://github.com/rust-lang/reference/pull/1630 -- [type-layout] Document bit validity of structs and padding - (Not really sure the status) https://github.com/rust-lang/reference/pull/1588 -- Clarify requires of Prefix/Directive validity requirements - Amanieu requested a change, but no response since August https://github.com/rust-lang/reference/pull/1248 -- Update object safety to match impl around self as receiver - Asked T-types for a review, but no response Rest of page 2 is mostly waiting-on-author: https://github.com/rust-lang/reference/pulls?page=2&q=is%3Apr+is%3Aopen ## Chat https://hackmd.io/ws8REEONQXCuoacMHzFFfA 8:03 avatar Josh Triplett Josh Triplett says: https://doc.rust-lang.org/std/ptr/index.html#provenance 8:26 Sid Askary Sid Askary says:Sorry, have to drop off. 8:30 me says:It's always the small ones that end up being some of the hardest. 8:36 me says:Excuse me. I will brb. 8:38 avatar Josh Triplett Josh Triplett says: https://ghc.gitlab.haskell.org/ghc/doc/users_guide/separate_compilation.html#orphan-modules-and-instance-declarations 8:47