Repository navigation
P264: rebind checked bus copies without a cold rebuild - #2007
Merged
trissim merged 3 commits intoOct 19, 2026
Merged
Conversation
There was a problem hiding this comment.
Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.
An organization owner can choose the repository organization as the billing source for reviews requested by members with a paid Copilot license. Learn about Copilot code review billing.
…ix/p264-private-bus-relocation
This was referenced Oct 19, 2026
trissim
added a commit
that referenced
this pull request
Oct 19, 2026
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
P264: concrete recovery records and chunked tail commits
Head:
787a99bfb20e2c0b87edafb5eb7cab989126f972, including #1996. This PR is stacked onintegrate/clean-batch-10.Owner repair
CheckpointSeal.recover,FinalSeal.recoverandPendingSeal.recoveralways return a prefix record. The nullable result and caller's None branch are deleted.PrefixWitnessowns the proved cursor and rebinds its physical revision before tail continuation.require_sourcestill checks root, size, tail and sequence. Shared reads escalate before a repair; the write lock owns all publication.CheckedPrefix.commit_tailowns the shared SQL/pending/final publication path used by ordinary appends, physical rebinding and recovery. Seals own the recovery end and chunk boundary. Existing WireLog/WireScan decode only the requested range and continue its prior sequence.Real grown-copy proof
The copied prefix and all appended bytes came from the actual live bus. No fabricated records, model input, extra agent, or uncertain input replay. A production WireScan profile counted only the real appended rows:
The new digest exactly matches the live source's complete saved digest.
p264-grown-copy.pyandp264-grown-copy.logare under the sole owned scratch directory/home/ts/.cache/agent-scratch/pi-cn-load/. Itsp264-real-bus/is a disposable copied-root diagnosis, not an active agent or installation. Combined checkout/scratch remains below 3 GB.Checks and remaining result
Source ratchet on this pushed head: rc=0; positive deltas {}; introduced {}. It ran alone in background under timeout 300 and faulthandler, using the shared QA interpreter and the exact existing pinned audit source. Tools ratchet: rc=0; positive deltas {}; introduced {}. The two checks ran serially. No suppression, cast, sentinel, new environment, clone or altered baseline.
The combined target was handed to claude-code for the actual installer rehearsal and live deployment. No full rehearse rc=0 or installation is claimed by this PR until that result arrives. Unit tests on a fake bus are not being substituted for that proof.
Assumed: the latest instruction to remove the cold bound supersedes the original keep-cap instruction; the private bus is append-only under its existing write lock; actual copied history is data, not permission to replay input; installer/build restrictions remain in force for this worker.
Agent-Comms-Order: 1d7ec808c13c