T-spec meeting 2025-05-22

--- title: T-spec meeting 2025-05-22 tags: ["T-spec", "meeting", "minutes"] date: 2025-05-22 discussion: https://rust-lang.zulipchat.com/#narrow/channel/399173-t-spec/topic/Meeting.202025-05-22/ url: https://hackmd.io/hIZsj4_TTHuhgHq90-ykDw --- Attendees: - Joel Marcey - Pete LeVasseur - Pierre-Emmanuel Patry - Eric Huss - TC - Josh - Jack Regrets: - Niko Matsakis Agenda: - Updates to the agenda? - Recap from spec discussions at Rust Week - Reference Issues/PR review - Should we start looking at FLS issues and PRs too? ## Updates to Agenda ### FLS and Reference Pete: Harmonizing the FLS and Reference. Has that started? Eric: I've done a chapter mapping between the two. All: 👏 Pete: I'm willing to help here. TC: One thing you could do would be to go through the FLS and see if there is anything there that should be in the Reference. Pete: Best format to do this type of work? TC: GitHub issues and PRs. Pete: Can I use Eric's mapping? Eric: It's on physical paper right now. Will try to get that in a form to share. TC: It's also helpful to know when a chapter in the Reference contains all the material in the matching section of the FLS. So going through and signing off on that relative completeness is also something you could do. ## Spec Week Recap https://hackmd.io/@davidtwco/B1JU3iNZxl Joel: We did office hours, but that was kind of a dud; few showed up. We had an officially scheduled meeting; Niko might have scheduled it, but wasn't there. Josh was there with me; Pete wasn't there. David Wood took some notes. There was generally agreement about the current plan about keeping the FLS out of project processes. There was curiousity about whether the spec team should own other documents like the UCG or the nomicon. Josh: General feeling in the room at the meeting was there was a warm fuzzy feeling that maybe the spec team should consider whether owning the rustnomicon makes sense. Josh: Split out the nomicon to see where it might make sense to add to the spec, particularly around unsafe. TC: Who was there? Joel: Florian was there, along with David Wood, Josh, and myself. Josh: I don't recall who else was there as part of the Project, but there were many other folks there who were interested in the spec. TC: The nomicon is primarily an informative document and the normative material actually goes in the Reference. The normative material from the UCG is already included in the Reference (per its own README). The residue are things that did not achieve consensus. Joel: Ferrous to upstream their changes to the Rust Project first. Seeing the Project FLS as the authorative document. Josh: Also Ferrous to potentially provide resources to help with spec writing, which is nice. Joel: What the means in practice remains to be seen when we actually have real work to be done. TC: Having looked at some FLS PRs, we don't have the context on the document at the moment to be approving these ourselves if they're not coming from an existing author. So we are pinging Lukas Wirth about these. Maybe Sam should be involved as well. Joel: Dual approval for FLS PRs? TC: Makes sense. Joel: We also talked about documenting the borrow checker. Josh: Full documentation of the borrow checker vs. minimum bar of the borrow checker. Josh: If we have fully specified the borrow checker in 1.0, then we would not have been able to document NLL without breaking the spec. ## Reference Issues/PR Review ### How to handle pre-stabilization? Eric: There's a problem with stabilization PRs, where we could review the documentation and "approve" it in a sense that it looks correct at that point in time. However, during the stabilization process sometimes PRs change the behavior, and it is difficult to follow those changes. Very recent example is AVX512 https://github.com/rust-lang/reference/pull/1778, where the behavior changed during stabilization. And the two `cfg(version)` and `if let` guards are undergoing changes right now. Josh: Could use GitHub request review change. Eric: The problem is more on the stabilization PR, and probably don't want to put an unconditional "hold" on every PR like that. ### New Issues since last Meeting - [Named Local Variable vs Local Variable](https://github.com/rust-lang/reference/issues/1832) - temporaries might be a distinction here - May not be worth spending time on these types of issues. - Maybe use GitHub labels for priorities or difficulty. ### New PRs since last meeting - [Add doc for keylocker target features](https://github.com/rust-lang/reference/pull/1829) - Waiting for stabilization. Eric has reviewed. Remove waiting on review. - [Add doc for `sha512`, `sm3` and `sm4` target features](https://github.com/rust-lang/reference/pull/1830) - [Document `#[cfg(version(...))]`](https://github.com/rust-lang/reference/pull/1828) - Amount of discussion currently happening on stablization PRs are huge. A huge amount of volume. - [`if let` guards documentation](https://github.com/rust-lang/reference/pull/1823) - Early in the process. ## Reviewing FLS Issues and PRs Joel: Should we have time dedicated in our spec meetings to review FLS issues and PRs? Pete: If we can get Lukas' (or maybe Sam's) time to review during the meeting, that would make sense. TC: We won't make a lot of progress of FLS PRs without Ferrous' help right now. TC: We can look through the queue and see what Lukas has approved. Joel: Ask Ferrous folks (Lukas and Sam) if they would be willing to attend some cadence of meetings to help us with the review process for the FLS. TC: Make sure they know we are available to help them get what they need here. ## Jitsi Chat Meeting agenda and notes: https://hackmd.io/hIZsj4_TTHuhgHq90-ykDw 9:59 me says:Rust Week spec discussion: https://hackmd.io/@davidtwco/B1JU3iNZxl 9:59 me says:Just randomly trying to remember folks at the meeting.... I believe tshepang might have been at the meeting. 10:14 PL Pete LeVasseur Pete LeVasseur says:😀 10:19 J Josh Josh says:👏 10:36 I have to step away for two minutes. Brb. If someone can take any important notes on the discussion, I would appreciate it. 10:44