T-spec meeting 2025-05-08

--- title: T-spec meeting 2025-05-08 tags: ["T-spec", "meeting", "minutes"] date: 2025-05-08 discussion: https://rust-lang.zulipchat.com/#narrow/channel/399173-t-spec/topic/Meeting.202025-05-08/ url: https://hackmd.io/opHOmKTxRwGhe1AOiCAuXg --- Attendees: - Joel Marcey - Eric Huss - TC - Josh Triplett - Jack Huey - Pete LeVasseur Regrets: - Niko Matsakis Agenda: - Updates to the agenda? - FLS is now managed by the Rust Project - Integrating it into the Project build process - Spec team at Rust Week - New Reference Issues/PRs - No meeting next week (Rust Week) ## Updates to the Agenda https://github.com/rust-lang/rust/pull/137323#issuecomment-2861605847 ## Next steps for the FLS https://github.com/rust-lang/fls Joel: The FLS is now part of the Project. I'm interested now in how we get this published as part of the build processes of the Project. TC: It is already rendered at https://rust-lang.github.io/fls. So in some way it is already part of the Project documentation processes. TC: But we could do it insted like the Reference. Pros - it is like the Reference. Eric can maybe speak to how to do that. We'd submodule it, etc. Cons - while it has more custom link, gets included into the tarballs, etc., the rendered view then lags as compared to the latest changes. Eric: How would we handle versioning? e.g., reference 1.84 ships with Rust 1.84. Would the same thing happen for the FLS? The Book is behind, but specifies at the beginning which version it is for. FLS could be similar? Ferrous publishes multiple versions of the FLS, but it is private behind a login. The public docs only show the [latest main](https://public-docs.ferrocene.dev/main/). Joel: Will the FLS reach parity to the release versions similar to the Reference? Josh: Two ways that could happen - FLS work speeds up substantially such that it keeps up with the language, or the language development slows down (hopefully never). TC: Over the long term, going chapter by chapter and getting down to one document over time is how I'd imagine we'd reconcile that. (Discussion to the effect of how we'd get to a point where the FLS could document the latest Rust release and be included in the bundle.) TC: Maybe folks in the Safety Critical Rust Consortium can become members of the Rust Foundation and help contribute so that the Foundation could hire a writer. Pete: Agree with this as a possible option. Joel: Custom domain for the FLS? TC: No; other Project documents that aren't published as part of the Rust release just end up published on `github.io`, as this is. E.g., the rendered RFC repo. Eric: We can create versions of the FLS that people can refer to. Basically past snapshots. Jack: How does this work in the context of LTS? (Discussion about that.) TC: Separately, an option for versioned releases is to push them to GitHub Releases. TC: Notably, also the FLS repo doesn't have any historical version tags. Joel: Will ask Ferrous if they snapshot versions and publish those. TC: If we want to publish multiple rendered versions on the web, we should just roll it into the build process for Rust, like the Reference. Conclusion: https://rust-lang.github.io/fls/index.html will be published for now (from latest main). If we learn more about people needing versioned snapshots, then we will investigate more about publishing in rust's build system (like https://doc.rust-lang.org/stable/). If we decide to publish on doc.rust-lang.org, we discussed the concerns about the potential mismatch of versions between released snapshots like https://doc.rust-lang.org/1.84.0/ and the FLS at https://doc.rust-lang.org/1.84.0/fls/ being a different version. There was some disagreement of whether or not that will be a problem. Currently the FLS specifies which version it is for, but Josh did not like the idea of people getting confused despite that warning. Eric mentioned that the Book does this (and lags behind), and although sometimes there are people who file issues about that, it generally hasn't been a problem. Generally we leaned toward accepting that disconnect (I think?), and also looking into finding resources to ensure the FLS doesn't lag. That may need more input from interested parties to see what kind of concerns they may have. ## Spec team at Rust Week @JoelMarcey and @joshtriplett are holding spec office hours on the afternoon of Friday 16 May. We don't have an officially planned presentation or agenda, but some items that we were considering for people to come talk to us about are: 1. Have an open forum on how people would like to see the FLS and the Reference live together as part of an overall Rust specification. 2. Continue working on the FLS integration into the Project's build processes 3. Solicit ideas of what people would like to see in a specification overall. 4. Co-evolution processes for the FLS/Reference and Rust 5. Integrating the specification into the RFC processes, including improvements to the RFC framework overall to support 6. From Pete: Coding Guidelines in Safety Critical Rust Consortium 7. Talk about what you want ## New Reference Issues and PRs ### Issues - [1805 - Simplify syntax diagrams](https://github.com/rust-lang/reference/issues/1805) - [1806 - Reference is incorrect about macro expanding expressions in builtin attributes](https://github.com/rust-lang/reference/issues/1806) - [1808 - Document Restrictions::NO_STRUCT_LITERAL better](https://github.com/rust-lang/reference/issues/1808) - [1809: Document Restrictions:STMT_EXPR better](https://github.com/rust-lang/reference/issues/1809) - [1810 - Figure out how to better define let chain grammar](https://github.com/rust-lang/reference/issues/1811) ### PRs - [1813 - Add new temporary lifetime extension rule](https://github.com/rust-lang/reference/pull/1813) - [1816 - Minor fixes to $crate behavior](https://github.com/rust-lang/reference/pull/1816) - [1815 - Add a note that ! can only be coerced to sized types](https://github.com/rust-lang/reference/pull/1815) - [1779 - Mention the temporary scope of while let](https://github.com/rust-lang/reference/pull/1779) ## Other https://github.com/rust-lang/rust/pull/137323#issuecomment-2861605847 ## Chat https://hackmd.io/opHOmKTxRwGhe1AOiCAuXg 8:02 T TC TC says:👏 8:04 J Jack Jack says:I wonder if this conversation also makes sense in the context of a potential LTS type thing.