Repository navigation
Owner decision sheet D332+ (successor to #787): open rows and execution queue #1234
Description
Successor to #787, opened 2026-10-13 under F4a clause 4. #787 ran from D1 to D331 and is now closed as completed: every row on it is ruled, moot or struck. It stays the evidence; nothing there is re-asked here.
⚠ The rulings ledger no longer exists
The owner's workstation was wiped (owner, 2026-10-13). The local ledger MEMORY-RULINGS.md and its 2026-10-05 snapshot were on it, so this issue is now the only rulings register.
- IDs: next free id = max(id posted on this issue or Owner decision sheet D1–D293: one answerable place for #637 + #715 + #709 + #658 (superseded by successor — see latest comment) #787) + 1. Owner decision sheet D1–D293: one answerable place for #637 + #715 + #709 + #658 (superseded by successor — see latest comment) #787 ends at D331; D332 is booked below; D333–D334 are booked in the comments, so the next free id is D335.
- Ledger-only rows that were never posted:
- D321 and D322 survive because they are quoted in
hyperpolymath/julia-professional-registrydocs/decisions/0003-package-repos-in-metadatastician.adoc. D321 (2026-10-07): the Marid Tier-1 packages register in the professional registry now, and one owner-chosen package goes to General later. This was executed by registry PR Wave 4: Nix-mirror retirement (betlangiser…cloudguard-server) #109, and amended on 2026-10-13: all nine also go to General (see A6). D322: their UUIDs were re-minted as UUIDv8 profile T before registration. - D297 and D315 are lost. A code search of the hyperpolymath estate finds neither. If either matters, it has to be asked again.
- D321 and D322 survive because they are quoted in
- Anything else that lived only on that workstation is gone too, for example uncommitted worktree files of the D41 kind. A session that relied on local state should re-measure before acting.
Conventions carried over from #787
- F4a: a ruling is closed on the surface it governs, and a pointer comment lands here.
- Basis is stated on every ruling: owner direct answer (selector) or applied-unless-struck (D113).
- D296: question options name actors in the third person ("Jonathan enables…", "Claude opens…").
A. Execution queue: ruled on #787, not yet executed
| # | row | what is owed | acceptance |
|---|---|---|---|
| A1 | D16 | Remove ruleset bypass for app 1561 (Probot Settings, measured 2026-10-13) and app 946600 (not installed on the hyperpolymath account) wherever they still appear. D16's rule: an app that never pushes to the default branch loses bypass. | A census of bypass_actors over every hyperpolymath repo ruleset and every metadatastician org/repo ruleset shows neither id. |
| A2 | D14 | ClusterFuzzLite cflite_* stays only on Rust/Zig repos with real fuzz targets and is deleted where it runs template stubs. |
A census lists every repo carrying cflite_* with a has-real-targets verdict; deletion PRs are merged for the rest. |
| A3 | D24 | Pages across the 48 repos: keep Pages where a repo publishes a real website, and delete docs-only Pages workflows (docs go to BerryWiki, D295). | A per-repo classification table; deletion PRs merged; the list of real-site repos handed to Jonathan to enable Pages (API/UI-only). |
| A4 | D97 | Census of transfer candidates with per-repo cost (package-registry URLs, archived status, URL consumers), for a later ruling. | A table posted here, followed by a ruling row. |
| A5 | D298 | Revert metadatastician Branch-Floor 23868655 and Tag-Floor 23868851 to canon (deletion + non_fast_forward, bypass_actors: []). Both added rules leave in the same PUT (#787 comment 65). Measured 2026-10-13: Branch-Floor still carries 4 branch rules. |
Needs admin:org, a posted blast-radius list and a ?dry-run PUT. After it lands, GET shows exactly the two canon rules on each. |
| A6 | D67 (rev. 2026-10-13, amends D321) | The 9 Marid Tier-1 packages (MaridArango, MaridCodec, MaridControl, MaridCRDT, MaridIR, MaridLive, MaridRaft, MaridStorage, MaridTransport) also go to General now. They stay in julia-professional-registry with the same UUIDs. The local RegistryCI AutoMerge guideline check runs first (the guideline source is AutoMerge/src/guidelines.jl; RegistryCI.jl is a monorepo). |
The AutoMerge guidelines pass locally for all 9, including on the UUIDv8 profile-T UUIDs from D322. Then 9 General registrations are merged, or each blocker is posted here with its guideline text. |
| A7 | D330 | canonical-ums deletion: the online and local copies are already gone (see the close-out on #787). What remains is for Jonathan to glance at gitlab.com → Projects, since a private GitLab copy cannot be seen anonymously, and delete it if one exists. | Jonathan confirms that no GitLab copy exists. |
B. Open decisions
| row | question | (E) |
|---|---|---|
| D304 (open since 10-06) | Which host takes the dedicated encrypted, off-GitHub store for mirror bundles? D303 requires that bundles survive whole-account loss, and the workstation wipe is a live instance of the risk D303 names. | none stated yet |
| D332 (new, measured 2026-10-13) | D13's premise is false: the Probot Settings app IS installed on hyperpolymath (installation 145157322, app id 1561). D13(E) was "confirm no install, then delete the dead .github/settings.yml files", so its precondition fails and those files are live config wherever the app has access. Keep the app and its settings.yml files, or uninstall it and then delete the files? Rulesets already own branch protection, so the two write the same settings. |
Uninstall, then delete the files: one owner of repo settings (rulesets), and no app re-applying a second copy. |
C. Index back to #787
Final rulings and the close-out: #787 (last comment). Rows D1–D323 are in its body and comments; D324–D331 are in the close-out comment.
Activity
chore: rename kcq to kyq (Kubernetes YAML Query) (#9)
docs: kcq is now kyq (#5)
docs: kcX is now kyX (#1237)
hyperpolymath commentedon Oct 13, 2026
D333–D334: kcq and kcX renamed (owner, 2026-10-14)
ID base: max(this issue D332) + 1 = D333.
kyqis taken by an empty 0.0.0 placeholder from 2021, so any bun/npm publish uses a scope (@metadatastician/kyq).metadatastician/kcqredirects (301). In-repo references: metadatastician/kyq#9 (kyq_chora.deed,kyq-launcher.sh). Consumer docs: metadatastician/affinescript-libraries#5.:repo-uuidkeeps its preimagegithub.com/metadatastician/kcq: a name-based UUID identifies the repository, and a rename does not mint a new one (ADR-008; aliasing remains #1230's question).nostosvocabulary, AFFIRMATION, ADR-008,check-uuid-v8.shcomment), hyperpolymath/rsr-template-repo#234 (the template), the seven template-derived CLAUDE.md copies (autoformal-fidelity#2, heron-polyfeed-reader#2, resource-algebra#2, xmlpull-zig#2, zig-boringtls#2, zig-dnssec#2, zig-icu-collation#2), hyperpolymath/boj-server#364 and hyperpolymath/QuantumCircuit.jl#42.Not renamed, on purpose: dated records keep the name they were written with: affirmations "as of" a date (
docs/affirmations/AFFIRMATION-2026-10-07.adoc, boj-server's and kyq's AFFIRMATION), the closed #787 and the ruling texts of D309, D313 and D320. A2ML descriptiles are not hand-edited (D313).Owner follow-up: the GitLab and Bitbucket mirror repos, if they exist, are still named
metadatastician/kcq. The kyq deed's(forges …)now names themmetadatastician/kyq, so rename them on those forges.Next free id: D335.
🤖 Generated with Claude Code
docs: kcX is now kyX (#2)
docs: kcX is now kyX (#2)
docs: kcX is now kyX (#2)
docs: kcX is now kyX (#2)
docs: kcX is now kyX (#42)
docs: kcX is now kyX (#234)
docs: kcX is now kyX (#2)
docs: kcX is now kyX (#2)
docs: kcX is now kyX (#2)
docs: kcX is now kyX (#364)
docs(reorg): standards tidy-up deep dive, 220 issue dispositions, bot…
hyperpolymath commentedon Oct 14, 2026
D304 and D332 ruled; A7 closed (owner, 2026-10-14, selector)
Basis: owner direct answer (selector). Three questions were put through the selection UI in session
01WbjuQvcerJjDjwqTVhomNf, options worded per D296. This comment books no new id; the next free id stays D335..github/settings.ymlin the estate, one PR per repo. Rulesets are the single owner of repo settings (D94–D96) and nothing re-applies a second copy. D13's (E) arm is kept with its precondition inverted..github/settings.ymlcarriers is running in this session. The deletion PRs follow the uninstall, in that order, because the files are live config until the app is gone.Execution route for A1–A3 (owner answer, same session): Claude attaches every hyperpolymath repository to the cloud session one at a time and runs the
bypass_actorscensus and removal, thecflite_*census, and the Pages classification from there. The metadatastician org rulesets (A1's org half and all of A5) stay owner-side: the session proxy serves only repository-scoped endpoints, soadmin:orgwrites cannot be made from it.Owner follow-ups now owed: (1) provision the R2 bucket and token (D304); (2) uninstall the Settings app, then tell Claude so the
settings.ymldeletion PRs can open (D332); (3) theadmin:orgwrites for A1-org and A5 (D16, D298).Generated by Claude Code
docs(reorg): handover note for running the tidy-up packets from fresh…