agent: builder model: opus max_turns: 120 budget_usd: 30 —
keeps column for fn parameters is computed in the fixpoint (ADR-277 D17)Score: self-hosting lane; the build of an owner decision (#1012, option E, 2026-10-30) that unblocks K3a part 2 (k3a2-comptime-compute.md, next in the batch). Today’s rule (by value means kept) forces Arc and 'static captures on every callback through a trait. Step 1 of the To build list of #1012; steps 2 and 3 are b1012b-trait-column-refusals.md and b1012c-trait-column-data.md. Precondition: ADR-277 D17 and its Part I 6.5 / Part III text are on origin/develop (read D17 first).
crates/nikaia/src/contracts/keeps.rs infer gives a body-less trait method keeps for every parameter it takes by value (the Item::Trait arm, ~220-237), and the trait_keeps pre-pass (~144-171, 0.0.1312) copies that column onto every impl. D17: the column is the union, by position, of the default body’s keeps and those of every impl of the trait in its own package (private impls included, impls under tests/ not), computed inside the least fixpoint kept_by (crates/nikaia-std/src/tools/lends.nika); with neither it is empty (run). This step does it for parameters of fn type. Data parameters keep today’s by-value seed until step 3 (b1012c), so the hand-written Rust impls of the lowered host traits do not move yet.
nikaia run, 6929d69b)use std::collections
trait H {
fn ask(ref self, known: fn(ref String) -> i64? sync) -> i64? sync
}
struct Plain { base: i64 }
impl H for Plain {
fn ask(ref self, known: fn(ref String) -> i64? sync) -> i64? sync { return known("a") }
}
fn twice[T: H](host: ref T, names: ref collections::BTreeMap[String, i64]) -> i64 {
return host.ask(fn(n) { names.get(n) }) ?? 0
}
fn main() {
let mut names: collections::BTreeMap[String, i64] = collections::BTreeMap()
names.insert("a", 4)
let p = Plain { base: 1 }
println(twice(p, names))
println(names.len())
}Today: “Lifetime may not live long enough” at the lambda (the lowered ask takes Kept<dyn Fn>). Expected: 4, 1 (the free-function control prints them).
crates/nikaia/src/contracts/keeps.rs infer: the trait method becomes a node “H::m” of the graph. Its own evidence is its default body’s (else empty). Edge pairs (Passed { parameter, callee, at }), per fn parameter position: H::m -> X::m for each impl in the package (“the trait keeps what an impl keeps”) and X::m -> H::m back (the 0.0.1312 rule, moved inside the loop). Remove the trait_keeps pre-pass and the by-value arm for fn parameters; the impl-side union (~199-215) is then the back edge. The node and its impls are one SCC, so they agree by construction; the least fixpoint keeps a storage-free cycle run.keeps_its reads library.functions["H::m"]). The comparison and refusal are step 2’s.keeps_its falls back to kept for a callee absent from own (fail closed): every trait method, pub or not, needs its own entry, or a private trait reads kept.kept_args (check_core_args.nika) and the emitter (emit_function.nika ~565) read the column and need no change of their own; if one does, say so in Body:.tests/compiler/tests/ (+ .stdout), each also run against the unfixed compiler:
4, 1.Keeps { held: Vec[fn(i64) -> i64 sync] } pushes) and one that runs, through fn hand[T: H](mut h: T, n: i64) -> i64 { return h.take(fn(x) { x + n }) }: prints today and after; the column is kept, both impls lower Kept. Call the held ones with for f in k.held { println(f(10)) } (let first = k.held[0]; first(10) hits a separate defect, not yours).a_trait_method_callback_is_handed_on.nika still prints 6, 3, 9.impl Fn).relay[T: H]), no storage, called with a borrowing lambda: stays run, prints.tests/compiler/tests/commit_dry_run.nika (toolchain(), files into mktemp -d, process::run). (f): one program and the same with its items reversed give an identical nikaia.contracts through nikaia lower; print same. (i): a package whose src has a pub trait H and no impl, and whose tests/ file uses it with an impl that only runs: nikaia test --project passes (today rustc E0049: the library trait lowers Kept, the test impl impl Fn), and the package’s ledger shows no keeps for H::ask.lower-std fixpoint clean (docs/builder-facts.md): only files whose lowering moves. Commit contracts: a trait method keeps what its package's impls keep (#1012), Part of #1012.(a)-(f) and (i) pass through nikaia test (copy a file to the scratchpad to nikaia run it alone), the it filter contracts and the guard set of .claude/agents/builder.md are green, the fixpoint is clean; pushed. Changelog: one line (a callback through a trait method borrows its captures when no impl in the trait’s package keeps it).
An edit outside crates/nikaia/src/contracts/keeps.rs, crates/nikaia-std/src/tools/*.nika and their lowering (lowered/*.rs, lowered.rs index lines), and the new tests; a hand-written Rust impl of a lowered host trait has to change (that is step 3’s: Found: with the trait and the file); an error in another agent’s file; the trait’s impls of the package are not reachable from infer (Found: with where they are).
Not yours, never stage: every file git status shows modified or untracked at your start, and every docs/briefs/* file except your report.
docs/builder-facts.md first. develop in the main checkout; git status at start: other agents’ uncommitted files are not yours, never touch or stage them.git add <files> only; commit after each step; git push origin develop fast-forward, no rebase (rejected: report Pushed: rejected, go on committing).cargo fmt, cargo clippy --all-targets -- -D warnings, scripts/pre-push-check.sh, and the gate of .claude/agents/builder.md (#852).CHANGELOG.md, badges; no new NK code. New tests in Nikaia only; never the full suite. No foreground sleep, no polling; scripts in files.std or the toolchain lacks stops the work: no code around it unless an ADR or Part I-III refuses it (name it). Else Result: blocked, Found: gap + probe, Needs: owner decision (CLAUDE.md, A missing form stops the work).NIKAIA_CACHE_LIMIT=2G; at most two clones with target/ in the machine; delete your own clone’s target/ and target/release/nikaia-scratch at the end; report GB freed.Pushed: line names the pushed sha. A report without a Pushed: line, or with a placeholder left in (TC, DISK), is not done and cuts the batch (k3a1 2026-10-29: Result: done, nothing committed)./home/user/Nikaia/docs/briefs/b1012a-trait-column-fn.md.report.md)Fields of docs/architect.md, Briefs and reports: Result: done|noop|partial|blocked (the bare word alone on that line: Result: done, never a summary after it, which cuts the batch, b869a 2026-10-27; Owner: is a question for the owner or none, never a role), Files, Commit, Body (optional, one paragraph for the commit message), Fixes, Pushed, Tests, Gate (green|red), Deviation, Found, Needs, Changelog, Owner, Disk freed, and for a move Rs2nika: (the kept N of M lines of rs2nika --score, then the three costliest hand edits; none for a task that moves no Rust, none (redesign) for a brief with Shape: redesign and no Convert: line). none where nothing, and then the bare word: Deviation: none (a note) cuts the batch (batch.nika compares exactly); a note goes into the commit message. Found: is only for a defect outside your change (it cuts the batch); a process event such as another session resetting or rewriting your files is no Found:, it goes under Needs: (m3a 2026-10-28); a converter draft that came out wrong is no Found:, it is one of the costliest edits under Rs2nika:, which the converter lane reads (e12-4b 2026-10-27); how you solved your own step goes into the commit message. Deviation: is only a step of the brief built otherwise than written or not built (a choice the brief left to you, such as keep what X needs and say so, is no deviation: it goes into the commit message; nor is building two steps in one commit when every step is built, e12-3a 2026-10-27, nor a step another commit on develop already built, shown by your test failing with that commit’s change reverted, b889 2026-10-27: name the commit in Body:, nor a test’s expected .stdout/.stderr taken from what the program really prints where the brief’s sketch of it was short or wrong, b934 2026-10-29: say so in Body:, nor a lowering that follows the ADR governing it where the brief’s expected lowering contradicts that ADR, b933 2026-10-29: name the ADR in Body:, nor re-keying an allow-list entry (KNOWN) whose hash changed for the same program and the same reason, b-fixforward-sweep2 2026-10-28: say so in Body:); how you verified (e.g. a test not run against the unfixed compiler) goes under Tests:. A noop says why under Tests: (what was probed), not under Deviation:. Final answer, one line: done <sha>, report <path> (or BLOCKED: ... / PUSH REJECTED ...).
Attribution: Co-Authored-By: Claude Opus 5.5 noreply@anthropic.com Claude-Session: https://claude.ai/code/session_01GFeMqxRbtUMiGVyDMxsvEq