Repository navigation
v2 self host frontier - #11928
Closed
briansrls wants to merge 12 commits into
Closed
v2 self host frontier#11928briansrls wants to merge 12 commits into
briansrls wants to merge 12 commits into
Conversation
Wind-down of the self-host frontier lane. Four deliverables that were live in pull requests, prepared locally, or found as defects are recorded as roadmap nodes so none of the state survives only in PR bodies and session messages. - self-host-service-family-on-literal-terminal - whole-service-witness-costs-more-than-its-budget - required-floor-passes-with-a-failed-phase - runner-throughput-witness-imports-absent-members ROADMAP.md is a projection of this authority and is not regenerated here: no compiler binary is built in this worktree and building one under the current load is the wrong trade. The generated-artifact heal step is the intended actuator for the projection. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Review 69209: two displaced_cost fields carried measurements with no named producer. DESIGN section 6 requires a measurement to be cited by naming the producer that re-derives it, never by copying numbers into prose, because a transcribed number is unreachable from the thing that owns it. Both now name the required floor's per-claim cost receipt — the required-floor-claim-cost artifact, which carries eval_steps per witness identity — as the producer, and date the reading. The compliant form already in this file (roadmap_authority.dag:1303, "from the run named in DESIGN.md Building & checks") is the pattern followed. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Ledger-Repair-Judged: docs/design-rung-drops.md Heal-Candidate-Run: 35520547733
The remaining work for the service family is prepared but unlanded. Name the branch and exact tip in the row's first_slice so whoever resumes finds the work rather than rebuilding it, and say plainly that it predates the 2026-09-20 merges and needs rebasing. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
…into roadmap-wind-down-items # Conflicts: # ROADMAP.md
Review 69295. Two displaced_cost fields still carried exact measurements alongside the producer. DESIGN section 6 says a measurement is cited by naming the producer that re-derives it, never by copying its numbers into prose, because a transcribed number is unreachable from the thing that owns it and rots without anyone touching either end. An earlier review read naming-beside-the-number as sufficient; this one reads the rule strictly. The strict reading is the safer one and costs little, so the exact figures are gone. What remains is the qualitative shape that makes each item legible - a unidirectional shift across a large minority of the population, and a witness several times over its budget - plus the required-floor-claim-cost receipt as the producer that re-derives both. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Review 69315, advisory finding. The row said the prepared service-family slice "needs rebasing before it is pushed", which was true when written and false by the time it landed in the diff: the branch was pushed to origin at 80a560d before its lane closed, precisely so the work would survive the lane. Say what is true now: the branch is on the shared remote so the citation resolves for anyone with the repository, and what it needs is rebasing onto current main before it can be proposed for landing. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Review 69328. The service-family row carried a dated staleness note in first_slice, which this module's own annotation above fn updated forbids in terms: the slice is consumed as "next executable slice" by current_state and the dispatch brief, so a position written there forks its meaning and rots with no time axis to say so. fn fields concatenates first_slice verbatim, so a worker would have been handed "rebase a 2026-09-20-stale branch" as the slice itself. first_slice is now the executable slice only, keeping the branch and tip citation. The staleness and the rebase requirement move to an updated(on:, note:) entry where a dated position belongs and where the page renders it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Reviews 69295, 69315, 69328 and 69339 all found the same blocker: ROADMAP.md is a committed projection of roadmap_authority.dag and diverged from it once four rows were added. A committed projection diverging from its authority stops the line, and it did. I had reported this as unfixable here, on the grounds that regeneration needs a gunbc binary: a local release build is the host churn the wind-down forbids, and a remote build is linux/amd64 on an arm64 container whose runners refuse `gunbc run` with HostBudgetUnreadable anyway. That was wrong about the available instruments. The required floor ALREADY performs the regeneration and uploads it: its log records `[file] write ROADMAP.md (68580 bytes)` and run 35529884525 carries a heal-repair-candidate artifact. This commit applies that artifact rather than hand-editing the projection. Verified by content hash at all three points the manifest pins: expected_head_sha 5cbc649 = HEAD base_blob e4d64cd = committed file proposed_blob e95d1c3 = applied file The derived focus count moves 132 -> 138 and the ci-control row required-floor-passes-with-a-failed-phase now renders as a bullet, which is what the reviews said must happen. Ledger-Repair-Judged: docs/design-rung-drops.md Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
…wind-down-items # Conflicts: # ROADMAP.md
Merging main moved roadmap_authority.dag, so the projection committed in the previous merge (taken from main's side to resolve a generated-file conflict) no longer matched the merged authority. The required floor regenerated it again; this applies that artifact rather than hand-editing. Verified by content hash at all three points run 35533100665's manifest pins: expected_head_sha c68eede = HEAD base_blob 395c145 = committed file proposed_blob 3b1048f = applied file Derived focus count is now 145; the ci-control row renders as a bullet. Ledger-Repair-Judged: docs/design-rung-drops.md Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
# Conflicts: # ROADMAP.md # dag/gunbc/roadmap/roadmap_authority.dag
Contributor
|
Closing: this PR is empty, and the conflict is a squash-merge artifact rather than unlanded work. #11853 squash-merged this branch's content into main at 02:43. Squash keeps the content and drops the history, so GitHub still diffs against the old merge-base and reports rows that are already on main as additions. Verified against current main ( and This also does not block other work: the follow-up #11933 is on its own branch off main and is CLEAN. — sent from proud-bat-569 |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Auto-opened by session-dashboard for session
proud-bat-569.Pushing to
roadmap-wind-down-itemsadvances this PR.Worker attestation
Before flipping this PR to ready for review, confirm each item:
npm test,cargo test) and the result.Closes #Ndirective.Summary
TODO: replace this paragraph with one or two sentences naming the change and its motivation. Reviewers read this first.
Test plan