Repository navigation
Reuse exact reader quote data under current snapshot consent - #26
Merged
Merged
Conversation
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Move the simulated old authority latency from 200ms to 250ms so the finite 30s budget expires before the historical checkpoint. Keep the strict indeterminate journal, zero result and failed disclosure assertions instead of racing a completed checkpoint at exactly 150 checks. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The read-only full Host/Runtime shadow harness exposed a Reply cache miss hidden by the earlier fixture: Host reader presentation requests carry null draft-data consent, while compound Reply snapshots carry the active draft consent. Their otherwise identical presentation keys differed, causing another MIME read and fresh pre-provider round trip.
After proving its own non-null exact expected consent, a compound snapshot can now look up the reader's cached presentation using null reader consent id/digest. This reuses only parsed source data. Full scope, connection/session revision, account, policy digest, client fingerprint, custody generation, source reference and durable stamp remain identical. Current source ownership and every fresh Host/session/custody/source/destination/expiry check still run; the ordinary cache key and first-write path are unchanged. No permission or consent is extended.
The compound fixture now uses the actual Host null-reader/non-null-snapshot envelopes, with no forced authority equality. True cold coverage restarts the Runtime. Baseline RED reproduced four authority callbacks instead of three. GREEN covers all ten focused Reply tests, including cached-data denial without snapshot consent, policy and connection-revision cache partitioning, first-save source pins, destination races and native saved-record reauth compatibility. At 350ms authority latency, warm plain/HTML/attachment modes measured 1.13-1.23s with three fresh checks and zero MIME reads; cold reads measured 1.56s with four checks and one MIME read. IMAP authority/data tests also passed 5/5. Independent static review and unchanged-line EOL audit passed.
Email29/App10/Runtime27 and Host137 requirements stay unchanged. Previous unpublished staged artifacts are superseded and must be rebuilt from the final reviewed merge. Hold merge/publication for the coordinator and full shadow/visible acceptance. No Send, tag, release, deployment, activation or live mailbox change was performed.
The initial CI attempt also reproduced a pre-existing historical observation fixture race: 150 simulated 200ms authority checks landed exactly on its 30s deadline, allowing a completed journal before disclosure was denied. The coordinator approved changing only its two hypothetical clock-cost literals to 250ms so expiry occurs before checkpoint. Strict indeterminate-journal, zero-result and no-disclosure assertions remain. Five isolated repetitions and all 17 observation-latency tests passed; no production observation behavior or assertion was loosened.
🤖 Generated with Claude Code