Repository navigation
Finish RSR-template instantiation + reconcile stack-diagram vs KRL-stack grounding #56
Description
Source: AFFIRMATION.adoc (2026-06-19 real-gate run).
State (verified 2026-06-19, commit 102499d)
The most rigorous repo in the estate (OCaml compiler + Rust LSP/WASM + Lean-4 type-safety proofs; TG-3 LANDED). MUST/TRUST pass, but:
- 5
{{PLACEHOLDER}}files + 8rsr-template-reporeferences remain. - The maintainer-confirmed README stack diagram (
KRL → TangleIR → VerisimCore → Skein/QuandleDB) dropped "KRL … not merely a query language" and softened the Skein/QuandleDB labels — a mild tension with the groundings that still emphasise those.
Acceptance criteria
- Clear the 5
{{PLACEHOLDER}}files + 8rsr-template-reporeferences. - Decide + document whether the simplified stack diagram supersedes, or should re-state, the "not merely a query language" / "Skein = engine" nuance (reconcile with the krl & Skein groundings).
Cross-references
AFFIRMATION.adoc(this repo); the groundings it tensions with: docs(grounding): reconcile Skein role to KRL stack — compute/backend engine Skein.jl#20, docs(krl): ground KRL identity + 5-role stack per canonical spec krl#35.
Out of scope
- Re-running the Lean/OCaml proofs (need
lean/dune; absent this session — see AFFIRMATION). - The VerisimCore placement itself — settled, maintainer-confirmed 2026-06-19.
Filed from the 2026-06-19 estate AFFIRMATION-rollout retrospective.
Activity
fix(governance): SPDX header + SHA pins in pages.yml, refresh stale s…
fix(rsr): finish template instantiation + enforce it (#56) (#90)
hyperpolymath commentedon Jul 29, 2026
Closed by PR #90.
Criterion 1 — template residue: cleared. The counts in this issue were low and the classification conflated two things. Re-measured: 8 files matched
{{...}}, 21 lines matched the template name — but most were meta-references (files describing the debt). Three would have been damaged by a naive replace, notablysrc/rust/src/eval.rs, where"...add{{}} block"is a Rustformat!escape, not a placeholder.The genuine residue is gone: all eight real
{{...}}tokens, the four contractiles whose subject was still the template (Intentfileasserted "This repository is the canonical template for RSR compliance"), anddocs/CITATIONS.adoc, which credited "Polymath, Hyper" for "RSR-template-repo" in all five citation formats.Criterion 2 — already satisfied by #78.
README.adoccarries the correction andAFFIRMATION.adochas "Erratum — 2026-07-21 / Correction 1 — the stack does not exist".Worth recording why: this criterion asked whether the simplified diagram supersedes or re-states the nuance. Both options were wrong —
KRL → TangleIR → …describes a dependency that never existed (TangleIRappears in no source file in either repo). It was withdrawn, not reconciled.Beyond the ask. All three QUICKSTARTs documented a dozen commands that do not exist (
just build,just test,just install,just run,just heal, …), asserted a config file the compiler never reads, and carried a multi-instance deployment section for a single stateless binary. Documentation that fails when followed is worse than none, so those are corrected and the worked example now shows real verified output.And the rule is now enforced.
methodology.a2mlhad always declaredreject-if-contains = [..., "rsr-template-repo"]with nothing enforcing it — which is exactly how the contractiles stayed mis-instantiated.scripts/check-rsr-instantiation.shnow enforces it in CI (justinstalled on the runner so the phantom-command half actually executes), verified adversarially in all four directions.