Repository navigation
chore: upgrade to template f63b965 (three-way merge) - #8
Conversation
Base is 8c7baf7 (the deed's :template-commit, tree verified), theirs is rsr-template-repo main f63b965, ours is sholes main. The deed records no mint answers, so only {{PROJECT_NAME}} (sholes) and {{OWNER}} (hyperpolymath) are filled, inferred from the repo's own files, and only in the files the upgrade adds them to raw. Owner-approved under the standing rule with limits: .cursorrules and .windsurfrules are deleted (the repo's copies equal the template's retired content with tokens filled), the stale build/docs-seed/ block is removed from .hypatia-ignore and the matching structural_drift entry from the rule-order policy. Also: the container-templates deed line is split onto three lines copied from the template, both container FROM lines are pinned by digest only, and LICENSE.adoc is added. The template's deploy.k9.ncl.in is not added. Co-Authored-By: Claude <noreply@anthropic.com>
Co-Authored-By: Claude <noreply@anthropic.com>
|
You are seeing this message because GitHub Code Scanning has recently been set up for this repository, or this pull request contains the workflow file for the Code Scanning tool. What Enabling Code Scanning Means:
For more information about GitHub Code Scanning, check out the documentation. |
The upgrade merge brought src/interface/Abi/Foreign.idr and Types.idr in with the raw lowercase project token, which repo-init would have filled at mint (build/just/repo-init.just, PROJECT_LOWER). Fill it with the value the Zig exports already use (sholes), so check-mint-tokens.sh passes. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0136P7Zm6f3TYUpUuzGd5KAQ
The template renamed template-{hunt,kennel,yard}.k9.ncl to .k9.ncl.in because a
template is not a component (rsr-template-repo#247). This repository kept the
old names, so the K9 validator rejected their TODO placeholders as unfilled
component fields (K9-S003, K9-S005). Rename only; content is unchanged and
matches the minted .in files in typesnake.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0136P7Zm6f3TYUpUuzGd5KAQ
|
Important Review skippedToo many files! This PR contains 251 files, which is 151 over the limit of 100. To get a review, reduce the PR to 100 files or fewer by splitting it into smaller PRs or changing its base branch. Upgrade to a paid plan to raise the limit. This review couldn't start because sufficient usage credits or metered capacity aren't available. Add credits or update usage-based reviews in the billing tab, then retry.
|
Values: SERVICE_NAME and PROJECT_NAME sholes, VERSION 0.1.0, PORT 8080, REGISTRY ghcr.io/hyperpolymath/sholes. The k9 validator (contract v1.0.0) returns rc=1 on the old file and rc=0 on this one. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0136P7Zm6f3TYUpUuzGd5KAQ
…entence) The deed ABNF has no escape for a raw carriage return in a string and the deed linter rejects it. Exactly one byte is removed; the words and the rest of the (anchor) clause are unchanged. Diff, shown with cat -A: - :one-sentence "Tired of phantom 40,000-line diffs and /usr/bin/env: 'bash^M' errors? ... + :one-sentence "Tired of phantom 40,000-line diffs and /usr/bin/env: 'bash' errors? ... File size drops by exactly 1 byte. This changes the anchor hash, so CLAUDE.md's generated region must be regenerated (held by the sweep session). Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0136P7Zm6f3TYUpUuzGd5KAQ
…anged Removing the stray CR in the (anchor) clause changed the anchor hash (ANCHOR@74f24421da8c to ANCHOR@2f9979d75e10). Regenerated with just claude-md. Only the two provenance lines change. Co-Authored-By: Claude <noreply@anthropic.com>
The template renamed its example K9 records to .k9.ncl.in. This file is identical to the template's apart from one project-name comment. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0136P7Zm6f3TYUpUuzGd5KAQ
The repo copy was in the old K9 format: pedigree.component_type missing (K9-S003, K9-N001) and process.spawn not listed in pedigree.security.capabilities (K9-S007). The template's copy, carried byte for byte by ordinal-notation, passes. No local change to it. Co-Authored-By: Claude <noreply@anthropic.com>
…ated region Copied verbatim from the template's CLAUDE.md (owner ruling 2026-10-26). It sits below the ARRIVAL-PACK:END marker, so just claude-md does not touch it. estate-rules asserts it in governance_policy_test.sh. Co-Authored-By: Claude <noreply@anthropic.com>
The canon identity block is now template-only (rsr-template-repo#378). Co-Authored-By: Claude <noreply@anthropic.com>
|



Summary
Upgrades sholes to rsr-template-repo main f63b965 by three-way merge. Base is 8c7baf7 (the deed's
:template-commit; its tree 7e5f626e equals the deed's:template-tree).The mint answers are INFERRED, not recorded. The deed has no
(mint-answers)clause. I inferred{{PROJECT_NAME}}=sholesand{{OWNER}}=hyperpolymathfrom the repo's own files. Applied to the base template, the inferred answers (names, owner, author, email, year, repo, forge) reproduce 73 of 132 token-bearing repo files byte for byte; with the service values read from the repo's owndeploy.k9.ncl(SERVICE_NAME sholes, VERSION 0.1.0, REGISTRY ghcr.io/hyperpolymath, PORT 8080) it is 82 of 132. The rest differ because of repo edits or tokens I cannot see. This is strong evidence, not proof. Only those two tokens are filled, only in files the upgrade adds them to raw:CHANGELOG.adoc,docs/reports/compliance/RSR-TEMPLATE-DAMAGE-2026-10-30.adoc,scripts/check-mint-tokens.sh,scripts/survey-template-damage.sh,tests/workflows/devcontainer_test.sh(its assertion"{{PROJECT_NAME}} (Podman)"now reads"sholes (Podman)", which matches this repo's.devcontainer/devcontainer.jsonname).Owner-approved (standing rule with limits)
.cursorrulesand.windsurfrulesdeleted. Check made first: each repo copy equals the template's retired content at the base with tokens filled (no project edits).build/docs-seed/block removed from.hypatia-ignore(the repo never had that directory), plus the matchingstructural_driftentry from.machine_readable/policies/hypatia-rule-order.json, as the mint does.Changes
sholes_chora.deedline 228 (thecontainer-templatesentry, 1159 octets) split onto three lines copied from template deed lines 223 to 225; identical to the repo's words modulo whitespace at the split. Deed pins moved to f63b965 and its tree.build/container/Containerfile: bothFROMlines pinned by digest only,cgr.dev/chainguard/wolfi-base@sha256:1f6b749929164c3d9e7cad0289254f39b2ba51e1dddaeb0fa57532ea03f2db31(AS builderkept). The digest was looked up by the sweep session on 14 Nov; it freezes the image and does not keep it fresh. The Containerfile comment andbuild/container/README.adocshow the digest-only form (Sonar ruledocker:S8431).LICENSE.adocadded (owner-approved option A text, two lines).docs/architecture/REPOSITORY-MAP.adocregenerated in a clean worktree. Lock files regenerated.tests/e2e/use_template_mint_test.shis the template's file unchanged (new in this repo; self-skips in a minted tree).Explicitly left out
build/container/stapeln/deploy.k9.ncl.inis NOT added, and sholes' owndeploy.k9.nclis kept exactly as it is. The owner's condition ("leave it out if the deed lists it removed") was not met: the deed's:removedlist is empty. But its premise, "the repo has no filled copy", was wrong: sholes hasdeploy.k9.ncl(5992 bytes), and diffed against the base template file the only differences are the filled tokens (sholes, 0.1.0, ghcr.io/hyperpolymath/sholes, port 8080). The template renamed that file todeploy.k9.ncl.in, an unfilled source the mint fills; a minted child carries the filled file and no.insource. The owner chose "leave it out"; the sweep session ruled option (a) on this evidence. Owner decision: veto if you want the.inadded..a2ml, anchor or:repo-uuidchange; no other licence edit..devcontainer/devcontainer.json,.machine_readable/ai/PLACEHOLDERS.adoc,.machine_readable/coaptation/witness-map.ncl,README.adoc,build/container/compose.yaml,build/container/stapeln/.gatekeeper.yaml,docs/status/PROOF-STATUS.adoc,build/container/stapeln/deploy.k9.ncl; the template deleted.github/CODE_OF_CONDUCT.md,.github/CONTRIBUTING.mdand.github/workflows/deed-validate.yml, and the repo had changed them, so they stay.RSR Quality Checklist
Gate exit codes read directly (upgraded tree):
src/interface/Abi/Foreign.idrandTypes.idrcarry raw{{project}}(repo files, untouched);tests/workflows/devcontainer_test.shis filledsholes_chora.deedcontains a raw carriage return (U+000D) inside a string at line 355. Pre-existing on main (main's deed also carries one CR). The deed edit approved here is only the line split, so I left it.Testing
Template test set: 51 pass on the upgraded tree vs 15 on main (the template adds tests). Fail on both:
julia_fmt,check_adoc_renders. Fail only on the upgraded tree:chain_mint,ffi_strings,kyaml_format,kyaml_ratchet,rhodibot_canary(tools missing in my sandbox: uuidgen, Zig, Go yq), and two real ones:tests/workflows/devcontainer_test.sh(new template test) fails: this repo's.devcontainer/devcontainer.jsonkeys are not sorted ("devcontainer: object keys must be sorted"). The file is the repo's; I kept it.tests/workflows/governance_policy_test.shfails: CLAUDE.md lacks the "Proofs" section the test asserts. Known owner-approved red; CLAUDE.md stays out of scope.CI is the arbiter.
Owner decisions
deploy.k9.ncl.inout (above)..devcontainer/devcontainer.jsonkey order: sort the keys (a small change to the repo's file) or leave the new test red.sholes_chora.deedline 355 (pre-existing): fix it, or leave it.{{project}}insrc/interface/Abi/Foreign.idrandTypes.idr: fill (value would besholes) or leave.No ruleset applies to
mainhere, so merge commits are fine.Generated by Claude Code
Update: K9-DEED conformance (read from CI on eabd14f)
K9-DEED conformancefails ("K9 validation failed for 6 file(s)"). The first errors are all on.machine_readable/arrival-pack/claude-md.k9.ncl:K9-S003(pedigree.component_type is missing),K9-S007(security flags requestprocess.spawn, which the grant does not list) andK9-N001(contract violation). I never touch the arrival-pack, so this is not fixed here. sholesmainshows no K9-DEED run (only abuildfailure), so this lane is new with the upgraded workflows. The arrival-pack K9 files in sholes need the owner (or the wave that owns the arrival-pack). The other failing checks on this head are the ones listed above: deed raw CR (decision 3), the three dead links (PR docs: fix three dead internal links #9), and mint-tokens on{{project}}in the Abi files (decision 4).Generated by Claude Code