feat: consolidate TraceDecay V2 PR8-PR13 delivery - #707
feat: consolidate TraceDecay V2 PR8-PR13 delivery#707ScriptedAlchemy wants to merge 1941 commits into
Conversation
🦋 Changeset detectedLatest commit: b4cf191 The changes in this PR will be included in the next version bump. Not sure what this means? Click here to learn what changesets are. Click here if you're a maintainer who wants to add another changeset to this PR |
|
Post-reopen correction pushed in a23f86a: source-neutral background reconcile no longer advances the scheduler epoch and cancel in-flight text activation; mounted hook overflow uses the explicit invalidating path. Evidence: RED reproduced 0/1, GREEN 1/1 for ordinary_background_reconcile_does_not_supersede_in_flight_text_work; cargo check --lib --locked passed. |
💡 Codex Reviewtracedecay/.github/workflows/release.yml Lines 408 to 411 in fc20c4f For Linux targets, this archive now contains tracedecay/release-please-config.json Lines 22 to 25 in fc20c4f The new npm publication job derives its version from the packed tracedecay/dashboard/src/workspaces/delivery/DeliveryPage.tsx Lines 80 to 82 in fc20c4f When the scope bar selects a project other than the active project, this query keeps the same cache key and continues fetching the unprefixed active-project route. The dashboard already provides tracedecay/src/tracedecay/lifecycle/registry.rs Lines 118 to 122 in fc20c4f When AGENTS.md reference: AGENTS.md:L159-L161 tracedecay/plugin/cursor-native-extension/src/extension.ts Lines 187 to 189 in fc20c4f At activation or language-client restart, ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |
fix(clippy): clear the workspace clippy failures on #707
Removing the re-export shims left their module doc comments orphaned in `context` and `graph`. The rest are pre-existing lints that only fire under `--all-features`, which compiles the `test-transport` support surface as non-test code: - the registered-test-server constructor gets the file's existing explicit `expect_used` allow, since `lib.rs` denies it only outside `cfg(test)`; - `http_application_registry` is read by the `cfg(test)` capacity journey and only reads as dead under `test-transport` alone, so the allow is gated to `not(test)` rather than renamed to a discard; - the hotpath cadence helpers keep `#[inline(always)]` with an explicit allow, preserving the author's intent instead of silently downgrading it. The `daemon_suite` restart wait is now the original helper recovered from the PR #707 head rather than a reconstruction: it additionally proves the runtime readiness receipt authorizes the exact configuration the SDK selected.
The relocation fix that anchored this test above the package allocated a `PathBuf` only to hand out a reference, which `-D warnings` rejects as `unnecessary_to_owned` and which failed the Clippy gate on #707. `validate_requested_workload` takes `&Path`, and `CARGO_MANIFEST_DIR` is a literal, so `Path::new` yields a `&'static Path` with no allocation and no temporary to borrow from.
|
#733 landed Clippy-red. Helper branch (not merged, shared checkout untouched):
Clears the #733 Clippy denials:
Adopt when you want: I will not land this on #707. |
|
Clippy helper is now on the Codex worktrees (still not landed on this PR head):
Same change as |
|
Isolated post-#721 verification (worktree Green (non-vacuous) on merge
Not green
Clippy helper remains on this tree at |
Ports the missing pieces of commit b1e41b712 ("fix(build): land the
batching commit's missing consumers") onto #707: HookCompletedReadinessDistributions
moved into tracedecay-agent-hosts with pub(crate) fields, so
input_rows_processed, input_rows_dropped_at_cap, and events_considered
are no longer reachable outside the crate. Adds the three public
accessors.
source_event and input_rows_received (also added by the source commit)
are not ported: nothing on this branch reads them through an accessor
today - the root-crate benchmark still compares via
serde_json::to_value(&readiness_distributions)["field"], which does
not need them.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Co-authored-by: Zack Jackson <ScriptedAlchemy@users.noreply.github.com>
5d8f2a1 to
c7fcf3f
Compare
…d6ed chore(vendor): stacked vendor-only PR for #707 — generated vendor churn, land first
…rules Co-authored-by: Zack Jackson <ScriptedAlchemy@users.noreply.github.com>
Selective lexical classes paid ~300ms at generation scale because each query with a fuzzy bound ran ORDER BY term against hash-keyed term_ids. Load in-fuzzy terms unordered once per reader and skip the scan when no group has a positive bound.
Open-time section digests used interned term_id SQL against every file. Dispatch those queries on the recorded format revision so revision 10 serves instead of surfacing a raw missing-column error.
Keep measured no-win Bytes generations in replay form so publication stays within prior-tip latency while retaining compact correctness at scale.
…plan-reopened' into agent/td-707-publication-regression-sol
Context reused the callers complete-neighborhood budget, hydrated every physical edge, then kept 20 related symbols. That CPU walk ran on every call and erased search warmth. Truncate at a small page, cache snippet reads by path, and keep serve-stale coverage markers.
Adopt the merged engine revision that coalesces transaction WAL frames while preserving commit fsync and recovery semantics.
…plan-reopened' into agent/td-707-publication-regression-sol
Move deterministic sanitization and external payload staging ahead of the SQLite transaction while keeping rollback armed through commit.
Submit ordered write windows through one blocking hop while preserving the exact failed statement index and SQLite classification.
Batch snapshot-provider admissions while chaining each session cursor and replaying typed non-durable windows through the scalar path.
Use one daemon-wide four-thread parse pool while preserving discovery-order persistence and the two-pass catch-up admission bound.
Persist revision-7 manifests over normalized content-addressed file and evidence segments so one-file edits reuse unchanged extraction bytes while old monolithic revisions remain explicitly readable.
Verify revision-7 component digests through replay, mark shared segments across canonical and retained manifests before retention, and report physical segment growth in the efficiency scorecard.
Read partitioned lexical metadata without entering the full-generation decoder, keep component bytes inside durable history bounds, and update corruption and legacy-format coverage for the new physical layout.
The response-gate wait in tools/call dispatch never observed the request's cancellation signal, so a client cancel during a held gate hung the route; the gate acquisition now races the signal and returns the typed cancelled terminal. Pre-cancelled cooperative dispatches are admitted instead of refused so the invocation authority records the authoritative settlement (a cancellation that raced registration still reaches the application executor), while the pre-admission terminal stays plain cancelled because the commit CAS is already lost. After the peer half-closes, the RMCP transport now also closes once every accepted request has settled, so cancelling clients that read until daemon EOF are not stranded waiting for a full peer close.
The graph-authority probe asserted that every awaiting-graph handler fails closed on an absent query port, but tracedecay_changelog and tracedecay_run_affected_tests refused earlier on their documented pre-admission validation (changelog proves git evidence first, affected tests require an explicit manifest). Give the probe a committed git fixture and each handler's required request shape so the calls reach the graph admission step the test actually pins.
Measure the post-EOF settled-requests wait and count settled-close versus full-peer-close outcomes so connection-teardown stalls are attributable.
Every handshake refusal a served engine writes advertises binary_version(), which reads the registered product runtime. Tests that drive the engine without building a handshake first (the unparseable-handshake and wire-drift refusal tests) depended on another fixture in the same process registering it, and failed alone once version provenance moved behind the product-runtime registration. Register the canonical test fixture inside the shared engine factory so every engine test runs under the process contract the daemon requires.
tools/list uses the unnamed-tool default, not tracedecay_context's catalog 40s, so the drain bound must outlive the actually resolved ceiling.
Oversized workspace-folder count is a front-door InvalidRequest; keep proving the rejection never initializes the analyzer or mints session access.
Keep selector-only tools on the active retained owner while preserving canonical alias validation, and require the transport authority for the suite that exercises it.
Replacement review for #421 after its accidental merge was reverted from master in 52a9aab. This commit has the exact tree of #421 head d4c67d4. Do not merge until explicitly authorized.