Repository navigation
test: prove self-update keeps live compression state across the OLD->NEW hop (proof-coverage audit gap 1) - #2724
Merged
Conversation
…NEW hop (audit gap 1)
Contributor
📦 Built Package ArtifactBranch: Option A — Install this exact build (recommended)npm install -g billion-context@0.0.2956-pr.2724.e305fb9This exact version is immutable and reproducible. The rolling Option B — Download artifact
tar xzf billion-context-pr2724.tgz
npm install -g packageInstall instructions are refreshed on each push. |
Contributor
📦 Merged Build PublishedThis merge to master was published as npm install -g billion-context@0.0.253-master.65ea85b # this exact build
npm install -g billion-context@master # always the newest merged build
|
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Closes the first 🔴 gap from the cache/identity proof-coverage audit: the upgrade chain (self-update -> restart) had ZERO coverage for "a live session's fold state must survive the OLD->NEW hop". e2e-registry and e2e-release-canary never touch session state — the canary's fold smoke is actually a refusal receipt (args:{} -> parse:missing-content, ok:true with no block), so nothing anywhere proved persisted folds survive an update.
What the new test proves
tests/e2e/e2e-update-state.test.ts(gated by ACP_TEST_REGISTRY=1, ~19s, hermetic local verdaccio + fixture tarballs like e2e-registry):compresscalls with explicit ref ranges (m00003..m00004, m00005..m00006) — both folds land (blocks b1/b2 ACTIVE).metadata.biliVersion === 0.1.191— proving the file the NEW build will adopt was written by the OLD build.--versionflipped.[Compressed conversation section]carriers present, POST_TURN visible. Status endpoint resolves the same sessionId; same session file; blockIds prefix-equal (b1,b2, no rewrite); maxRef/tokensCompressed/messageCount monotonically non-decreasing;metadata.biliVersionflipped to NEW (adoption, not rewrite).Teeth (A/B verified)
Sabotage run with
BILI_PERSIST=0on the restarted proxy (state never rehydrated) fails at exactlythe pre-upgrade folds must still be ACTIVE: folded raw material must not re-enter the wire after the update hop. Reverted; green run passes.Harness notes (pitfalls encoded in comments)
args:{}compress is a refusal receipt — explicit{startId,endId,topic,summary}ranges are the real contract.preserveRecentMessages=5ANDpreserveRecentTokens=5000(tail expansion) both gate fold ranges — the corpus needs ~6K tokens of tail filler after the fold chapters.Scope
Test-only addition; no src changes, no config surface, no kernel changes. Registry-gated so CI cost is opt-in (same family as e2e-registry/e2e-release-canary).
Refs: proof-coverage audit (chat thread 2026-10-11); related lanes #2695/#2658 (host-compaction coverage in #2707/#2720).