Repository navigation
chore: remove leftover junk and fix stale tests - #3132
Open
ScriptedAlchemy wants to merge 10 commits into
Open
ScriptedAlchemy wants to merge 10 commits into
ScriptedAlchemy wants to merge 10 commits into
Conversation
- Drop the root dogfood-journey.md, a one-off 2026-09-15 run report with machine-local paths that nothing references. - Drop the .changeset/ notes: the repo has no changesets config or tooling and releases are cut by release-please from conventional commits, so nothing ever consumes them. - Remove the stale "retrigger" comment lines from the Hawk workflow and the reference to a weekly schedule it does not have. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01V8Xg2pQQcTiyVB8ndJc5oo
broker_group_commit_test.rs was never declared as a module, so its three broker-level group-commit tests never compiled or ran. Declare it beside the other crate test modules and bring its expectations in line with current behavior: directory syncs happen on Unix only, and reopening the spool publishes once to clear the reconciled append intent, so the second batch is measured with a fresh barrier counter. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01V8Xg2pQQcTiyVB8ndJc5oo
DiffContextResultV1 now requires `test_gate`, so the SDK decoder rejected the fixture and the freshness-injection test failed with "value.test_gate is required". Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01V8Xg2pQQcTiyVB8ndJc5oo
revise_host_record skipped any write whose byte_offset and file_id matched the current record, and callers default an absent record to zeros. An empty home's Kimi and Pi discovery frontiers and zero-unit coverage rows therefore never landed, so status reported those providers as unobserved forever. Every write stores mtime >= 1, so treat mtime 0 as the absent record and publish it once. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01V8Xg2pQQcTiyVB8ndJc5oo
diff_context now always carries the test-gate verdict and hotspot rows carry 90-day churn, so the exact-payload assertions failed on both fields. The fixture project is one commit, so each row's churn is 1. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01V8Xg2pQQcTiyVB8ndJc5oo
revise_host_record took the defaulted record and recovered "absent" from mtime == 0. Callers already hold get_parse_offset's Option, so pass it through: an absent record is written, an unchanged stored one is not. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01V8Xg2pQQcTiyVB8ndJc5oo
The golden Pi fixture gained a contentless user message, which capture refuses, so the malformed-file test's fixture has nine lines and session_info spans the last one. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01V8Xg2pQQcTiyVB8ndJc5oo
The hotspots markdown rendering lists churn, and an empty diff_context reports a passing test gate, like the absent-file case. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01V8Xg2pQQcTiyVB8ndJc5oo
The server answers an unknown tool with an isError tool result carrying the unknown_tool problem, as the MCP suites pin, so the Inspector CLI exits zero and the old nonzero-exit check always failed. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01V8Xg2pQQcTiyVB8ndJc5oo
…anup-dead-code-4tjb49
|
![devin-ai-integration[bot]](https://avatars.githubusercontent.com/in/811515?s=60&v=4){kind=small}
Owner
Author
|
Test Linux was cancelled when it hit its 240-minute job timeout. This PR didn't cause that. Master's own Bazel Linux lane does the same: dispatch run 37424361757 on the merged #3124 head also timed out at 240 minutes. That same run showed failures across the workspace that no file in this PR touches:
The fix is #3133 ( Clippy has been queued since 08:52 behind the repo-wide Generated by Claude Code |
This branch has not been deployed
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
2 participants
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.
Summary
dogfood-journey.md, stale.changeset/*.mdentries, and Hawk workflow retrigger notes.Motivation
A sweep for dead code and leftover files, plus a full test run, turned up an unwired test module, several expectations that no longer match current output, and one real sessions bug.
Changes
crates/tracedecay-host-admission:broker_group_commit_test.rswas never compiled; it is now wired in through#[path].crates/tracedecay-sessions/src/runtime/source.rs:revise_host_recordnow takes the stored record as anOption. A host with no stored record now gets one published on its first sweep, instead of being skipped. Kimi and Pi pass the stored frontier through.pi_tests.rs: the Pi fixture has nine lines, not eight. The contentless user message is refused, sosession_infois the next observation.graph_analysis_test.rs: expectations now include hotspotchurnand the diff-contexttest_gate, in both JSON and markdown output and for empty diffs.scripts/mcp-conformance-smoke.sh: an unknown tool now returns anisErrortool result carrying theunknown_toolproblem, which the MCP suites already pin. The Inspector CLI exits 0 for that result, so the old "exits nonzero" check could never pass. The smoke now asserts the typed refusal.Test plan
bazel build //...passes when the change crosses package boundariesLocal runs, all Cargo-based (Bazel can't fetch its archives in this environment):
graph_analysis_test::(mcp_suite): 71/71 pass.mcp_suiteandtransport_acceptance_suitechecks for status (the sessions fix), ast-grep and v2 mounts pass.scripts/mcp-conformance-smoke.shpasses all checks.Checklist
CHANGELOG.mdupdated (under[Unreleased]if no version bump).envfiles included🤖 Generated with Claude Code
https://claude.ai/code/session_01V8Xg2pQQcTiyVB8ndJc5oo
Generated by Claude Code