feat(rehearse): engine::run_rehearsal + shipper rehearse CLI (#97 PR 2) by EffortlessSteven · Pull Request #127 · EffortlessMetrics/shipper

EffortlessSteven · 2026-04-17T14:03:55Z

Summary

Phase-2 preflight execution — the code that makes #97's config flags actually DO something. PR 1 (#120) added `RehearsalConfig` + CLI flag plumbing; this PR:

publishes every crate in the plan to an alternate registry,
verifies visibility on that registry via the same `version_exists`
mechanism live publish uses,
emits a `RehearsalComplete { passed, registry, summary }` event to
`events.jsonl` so the outcome is auditable from the event log alone.

Exposed as `shipper rehearse`.

Key invariants

Rehearsal registry must resolve to an entry in `opts.registries`.
Fails clean with a fix-suggesting error otherwise.
Rehearsal registry must differ from the live target (`ws.plan.registry`).
Refuses otherwise — the whole point is a sandbox.
Sequential execution; stops at first failure.
Does NOT touch `state.json`. Rehearsal is a pre-publish proof, not
an execution. Event log only, so the regular publish flow sees a
clean state dir.
`--skip-rehearsal` honored with a loud warning; outcome still
returned so the eventual hard gate can detect it.

Scope — explicitly NOT in this PR

Hard gate wiring `run_publish` → refuse without a passing
rehearsal (Rehearsal registry: strengthen preflight from dry-run to actual publish + install proof #97 PR 3).
Install / smoke check against the rehearsal registry (PR 3 or 4
— needs a consumer-workspace fixture).
plan_id binding of rehearsal outcome (needed alongside the hard
gate; simpler to land together).
Parallel rehearsal (not planned — correctness > speed for a
once-per-release rehearsal).

Tests

6 unit tests in `engine::tests`, all passing:

errors cleanly with no rehearsal registry
errors cleanly when rehearsal == live target
errors cleanly when the named registry isn't configured
`--skip-rehearsal` returns immediately, no events written
happy path emits Started + PackagePublished + Complete(passed=true)
cargo-failure path emits PackageFailed + Complete(passed=false)

CLI snapshots regenerated to reflect the new `rehearse` subcommand in
help output.

Test plan

`cargo test -p shipper --lib run_rehearsal` — 6/6 pass
`cargo clippy --workspace --all-targets --all-features -- -D warnings` clean
`cargo fmt --all --check` clean
`shipper rehearse --help` surfaces the new subcommand correctly
CI multi-OS green

coderabbitai · 2026-04-17T14:04:03Z

Warning

Rate limit exceeded

@EffortlessSteven has exceeded the limit for the number of commits that can be reviewed per hour. Please wait 40 minutes and 2 seconds before requesting another review.

Your organization is not enrolled in usage-based pricing. Contact your admin to enable usage-based pricing to continue reviews beyond the rate limit, or try again in 40 minutes and 2 seconds.

⌛ How to resolve this issue?

After the wait time has elapsed, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

We recommend that you space out your commits to avoid hitting the rate limit.

🚦 How do rate limits work?

CodeRabbit enforces hourly rate limits for each developer per organization.

Our paid plans have higher rate limits than the trial, open-source and free plans. In all cases, we re-allow further reviews after a brief timeout.

Please see our FAQ for further information.

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro

Run ID: 4b036848-a704-434f-8508-3cc8fa3565b1

📥 Commits

Reviewing files that changed from the base of the PR and between 6e95b83 and 87c5604.

⛔ Files ignored due to path filters (24)

crates/shipper-cli/tests/snapshots/cli_snapshots__help_text.snap is excluded by !**/*.snap
crates/shipper-cli/tests/snapshots/cli_snapshots__no_subcommand_error.snap is excluded by !**/*.snap
crates/shipper-cli/tests/snapshots/e2e_expanded__help_root.snap is excluded by !**/*.snap
crates/shipper-config/src/runtime/snapshots/shipper_config__runtime__tests__policy_combination_tests__snapshot_balanced_policy_typical.snap is excluded by !**/*.snap
crates/shipper-config/src/runtime/snapshots/shipper_config__runtime__tests__policy_combination_tests__snapshot_fast_policy_typical.snap is excluded by !**/*.snap
crates/shipper-config/src/runtime/snapshots/shipper_config__runtime__tests__policy_combination_tests__snapshot_safe_policy_typical.snap is excluded by !**/*.snap
crates/shipper-config/src/runtime/snapshots/shipper_config__runtime__tests__snapshot_tests__snapshot_all_flags_enabled.snap is excluded by !**/*.snap
crates/shipper-config/src/runtime/snapshots/shipper_config__runtime__tests__snapshot_tests__snapshot_alternative_registry_only.snap is excluded by !**/*.snap
crates/shipper-config/src/runtime/snapshots/shipper_config__runtime__tests__snapshot_tests__snapshot_balanced_policy_with_partial_config.snap is excluded by !**/*.snap
crates/shipper-config/src/runtime/snapshots/shipper_config__runtime__tests__snapshot_tests__snapshot_default_conversion.snap is excluded by !**/*.snap
crates/shipper-config/src/runtime/snapshots/shipper_config__runtime__tests__snapshot_tests__snapshot_encryption_with_env_var.snap is excluded by !**/*.snap
crates/shipper-config/src/runtime/snapshots/shipper_config__runtime__tests__snapshot_tests__snapshot_fast_policy_no_verify.snap is excluded by !**/*.snap
crates/shipper-config/src/runtime/snapshots/shipper_config__runtime__tests__snapshot_tests__snapshot_full_readiness_config.snap is excluded by !**/*.snap
crates/shipper-config/src/runtime/snapshots/shipper_config__runtime__tests__snapshot_tests__snapshot_linear_retry_strategy.snap is excluded by !**/*.snap
crates/shipper-config/src/runtime/snapshots/shipper_config__runtime__tests__snapshot_tests__snapshot_parallel_heavy.snap is excluded by !**/*.snap
crates/shipper-config/src/runtime/snapshots/shipper_config__runtime__tests__snapshot_tests__snapshot_resume_from_set.snap is excluded by !**/*.snap
crates/shipper-config/src/runtime/snapshots/shipper_config__runtime__tests__snapshot_tests__snapshot_safe_policy_max_safety.snap is excluded by !**/*.snap
crates/shipper-config/src/runtime/snapshots/shipper_config__runtime__tests__snapshot_tests__snapshot_webhook_with_secret.snap is excluded by !**/*.snap
crates/shipper-config/src/runtime/snapshots/shipper_config__runtime__tests__snapshot_tests__snapshot_with_registries.snap is excluded by !**/*.snap
crates/shipper-config/src/snapshots/shipper_config__tests__edge_case_snapshots__edge_policy_balanced_runtime.snap is excluded by !**/*.snap
crates/shipper-config/src/snapshots/shipper_config__tests__edge_case_snapshots__edge_policy_fast_runtime.snap is excluded by !**/*.snap
crates/shipper-config/src/snapshots/shipper_config__tests__edge_case_snapshots__edge_policy_safe_runtime.snap is excluded by !**/*.snap
crates/shipper-config/src/snapshots/shipper_config__tests__snapshot_tests__merge_cli_overrides_file_values.snap is excluded by !**/*.snap
crates/shipper-types/src/snapshots/shipper_types__tests__debug_snapshots__runtime_options_debug_snapshot.snap is excluded by !**/*.snap

📒 Files selected for processing (8)

crates/shipper-cli/src/main.rs
crates/shipper-config/src/lib.rs
crates/shipper-config/src/runtime/mod.rs
crates/shipper-config/tests/config_runtime_bdd.rs
crates/shipper-types/src/lib.rs
crates/shipper/src/engine/mod.rs
crates/shipper/src/engine/parallel/tests.rs
crates/shipper/src/runtime/policy/mod.rs

✨ Finishing Touches

🧪 Generate unit tests (beta)

Create PR with unit tests
Commit unit tests in branch feat/97-rehearse-execution

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

gemini-code-assist

Code Review

This pull request implements the rehearse command, enabling a pre-publish verification step against alternate registries. The implementation includes the run_rehearsal engine, new event types for auditing, and CLI integration. Feedback suggests simplifying the CLI failure output to remove redundant information and adding a test case for the scenario where a package is published but fails the visibility check.

gemini-code-assist · 2026-04-17T14:09:17Z

+                println!(
+                    "rehearsal FAILED after {}/{} packages against '{}': {}",
+                    outcome.packages_published,
+                    outcome.packages_attempted,
+                    outcome.registry_name,
+                    outcome.summary
+                );


The output message for a failed rehearsal is a bit confusing. The outcome.summary field already contains a well-formatted message like "rehearsal stopped at 3/5: ...", which includes the total number of packages. The current println! adds after {}/{} packages which is redundant and can be misleading, as outcome.packages_attempted is not the total number of packages in the plan.

For example, if the 3rd package out of 5 fails, the output would be something like:
rehearsal FAILED after 2/3 packages against 'my-reg': rehearsal stopped at 3/5: ...

This is confusing. Consider simplifying the output to rely on the summary, which is already comprehensive.

Suggested change

println!(

"rehearsal FAILED after {}/{} packages against '{}': {}",

outcome.packages_published,

outcome.packages_attempted,

outcome.registry_name,

outcome.summary

);

println!(

"rehearsal FAILED against '{}': {}",

outcome.registry_name,

outcome.summary

);

gemini-code-assist · 2026-04-17T14:09:17Z

+    #[test]
+    #[serial]
+    fn run_rehearsal_happy_path_emits_started_published_complete_events() {


The tests for run_rehearsal cover the happy path and a cargo publish failure, which is great. However, there's another important failure path that isn't covered: when cargo publish succeeds but the post-publish visibility check (version_exists) fails.

This can happen in real-world scenarios with registry propagation delays. Adding a test for this case would improve the robustness of the test suite.

The test would involve:

Mocking cargo::cargo_publish to return a successful exit code.

Mocking the registry server to return a 404 for the version_exists check.

Asserting that run_rehearsal returns a RehearsalOutcome with passed: false.

Asserting that a RehearsalPackageFailed event is emitted with ErrorClass::Ambiguous.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 2c445abc07

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

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".

chatgpt-codex-connector · 2026-04-17T14:10:46Z

+        .registries
+        .iter()
+        .find(|r| r.name == rehearsal_name)
+        .cloned()


Resolve rehearsal registry from configured list by default

run_rehearsal only searches opts.registries for the rehearsal target, but build_runtime_options leaves opts.registries empty unless the operator also passes --registries or --all-registries (see crates/shipper-config/src/lib.rs around lines 815-857). This makes the documented [rehearsal] registry = "..." config path fail with “not configured” even when [[registries]] contains that name, so shipper rehearse is unusable in the default config-only flow.

Useful? React with 👍 / 👎.

chatgpt-codex-connector · 2026-04-17T14:10:46Z

+
+        // Post-publish visibility check on the rehearsal registry. Reuse
+        // `version_exists` — same mechanism live publish trusts.
+        if !rehearsal_client.version_exists(&p.name, &p.version)? {


Reuse readiness backoff for rehearsal visibility checks

This check performs a single version_exists call immediately after cargo publish and fails rehearsal if it returns false, but registry visibility is often eventually consistent. The normal publish path uses verify_published with is_version_visible_with_backoff to avoid these false negatives; without that retry window, rehearsals can fail spuriously in CI even when the package becomes visible seconds later.

Useful? React with 👍 / 👎.

… PR 3) Close the rehearsal loop. PR 2 gave us \`shipper rehearse\` that publishes to an alt registry and emits RehearsalComplete. This PR makes live publish refuse to run unless a fresh, passing rehearsal receipt exists for the current plan_id. ## What's new - **\`state::rehearsal\`** module: a \`rehearsal.json\` sidecar next to \`state.json\` / \`events.jsonl\`. Schema versioned (\`shipper.rehearsal.v1\`), atomic write (\`.tmp\` + rename + fsync), crash-safe. Small enough to be grep-readable. - **Event schema additive**: \`RehearsalStarted\` and \`RehearsalComplete\` now carry \`plan_id\`. The events.jsonl audit trail now has enough information to replay the gate's decision after the fact. This is additive on variants that only shipped with PR 2 (#127), so no migration needed. - **\`engine::enforce_rehearsal_gate\`** + call site at the top of \`run_publish\`. Decision tree: 1. \`opts.rehearsal_registry = None\` → dormant, publish proceeds. Unchanged behavior for every caller who hasn't opted in. 2. \`opts.rehearsal_skip = true\` → proceed with loud warning. No fake-passing receipt is synthesized; auditors see "no rehearsal ran for this plan_id" in events.jsonl. 3. No \`rehearsal.json\` → refuse with fix-suggesting error. 4. Receipt's \`plan_id\` != current plan_id → refuse (stale). 5. \`passed: false\` → refuse with reason. 6. Fresh passing receipt → proceed, log a confirmation info line. - **\`run_rehearsal\` persists the sidecar** on completion. Best-effort: if the sidecar write fails, the events log is still authoritative and the gate will (correctly) refuse, blocking publish — the safe default. ## Tests 13 new tests. All green locally, clippy clean, fmt clean: - 3 roundtrip/persistence tests in \`state::rehearsal\` - 6 gate-helper tests (dormant / skip / missing / stale / failing / fresh-pass) - 1 e2e test that \`run_publish\` actually calls the gate and bails when rehearsal is required but missing Total rehearsal+gate coverage: 17 unit tests. ## Scope — not in this PR - **Install / smoke check** against the rehearsal registry (PR 4 or follow-on). The current gate trusts the rehearsal's publish + visibility verification. - **CI recipe** for kellnr or equivalent sidecar (separate PR; docs work, not code).

codecov · 2026-04-17T22:15:52Z

Codecov Report

❌ Patch coverage is 92.24138% with 27 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
crates/shipper/src/engine/mod.rs	94.49%	17 Missing ⚠️
crates/shipper-cli/src/main.rs	30.76%	9 Missing ⚠️
crates/shipper-config/src/lib.rs	83.33%	1 Missing ⚠️

📢 Thoughts on this report? Let us know!

Phase-2 preflight execution. PR 1 (#120) added config + CLI flag plumbing for rehearsal registry; this PR makes the flags actually DO something — publish every crate in the plan to an alternate registry, verify visibility, and emit a RehearsalComplete event so the outcome is auditable from events.jsonl alone. - EventType variants: RehearsalStarted / RehearsalPackagePublished / RehearsalPackageFailed / RehearsalComplete - RuntimeOptions fields: rehearsal_registry, rehearsal_skip - ShipperConfig::build_runtime_options populates with CLI > config > None precedence - engine::run_rehearsal: sequential, stops at first failure, emits events, does not touch state.json (pre-publish proof, not execution), verifies post-publish visibility via the same mechanism live publish uses, refuses to rehearse against the live target registry - shipper rehearse CLI subcommand Also regenerates the matching CLI snapshots (help_text, help_root, no_subcommand_error).

… PR 3) Close the rehearsal loop. PR 2 gave us \`shipper rehearse\` that publishes to an alt registry and emits RehearsalComplete. This PR makes live publish refuse to run unless a fresh, passing rehearsal receipt exists for the current plan_id. ## What's new - **\`state::rehearsal\`** module: a \`rehearsal.json\` sidecar next to \`state.json\` / \`events.jsonl\`. Schema versioned (\`shipper.rehearsal.v1\`), atomic write (\`.tmp\` + rename + fsync), crash-safe. Small enough to be grep-readable. - **Event schema additive**: \`RehearsalStarted\` and \`RehearsalComplete\` now carry \`plan_id\`. The events.jsonl audit trail now has enough information to replay the gate's decision after the fact. This is additive on variants that only shipped with PR 2 (#127), so no migration needed. - **\`engine::enforce_rehearsal_gate\`** + call site at the top of \`run_publish\`. Decision tree: 1. \`opts.rehearsal_registry = None\` → dormant, publish proceeds. Unchanged behavior for every caller who hasn't opted in. 2. \`opts.rehearsal_skip = true\` → proceed with loud warning. No fake-passing receipt is synthesized; auditors see "no rehearsal ran for this plan_id" in events.jsonl. 3. No \`rehearsal.json\` → refuse with fix-suggesting error. 4. Receipt's \`plan_id\` != current plan_id → refuse (stale). 5. \`passed: false\` → refuse with reason. 6. Fresh passing receipt → proceed, log a confirmation info line. - **\`run_rehearsal\` persists the sidecar** on completion. Best-effort: if the sidecar write fails, the events log is still authoritative and the gate will (correctly) refuse, blocking publish — the safe default. ## Tests 13 new tests. All green locally, clippy clean, fmt clean: - 3 roundtrip/persistence tests in \`state::rehearsal\` - 6 gate-helper tests (dormant / skip / missing / stale / failing / fresh-pass) - 1 e2e test that \`run_publish\` actually calls the gate and bails when rehearsal is required but missing Total rehearsal+gate coverage: 17 unit tests. ## Scope — not in this PR - **Install / smoke check** against the rehearsal registry (PR 4 or follow-on). The current gate trusts the rehearsal's publish + visibility verification. - **CI recipe** for kellnr or equivalent sidecar (separate PR; docs work, not code).

… PR 3) (#133) * feat(rehearse): hard gate — run_publish requires passing rehearsal (#97 PR 3) Close the rehearsal loop. PR 2 gave us \`shipper rehearse\` that publishes to an alt registry and emits RehearsalComplete. This PR makes live publish refuse to run unless a fresh, passing rehearsal receipt exists for the current plan_id. ## What's new - **\`state::rehearsal\`** module: a \`rehearsal.json\` sidecar next to \`state.json\` / \`events.jsonl\`. Schema versioned (\`shipper.rehearsal.v1\`), atomic write (\`.tmp\` + rename + fsync), crash-safe. Small enough to be grep-readable. - **Event schema additive**: \`RehearsalStarted\` and \`RehearsalComplete\` now carry \`plan_id\`. The events.jsonl audit trail now has enough information to replay the gate's decision after the fact. This is additive on variants that only shipped with PR 2 (#127), so no migration needed. - **\`engine::enforce_rehearsal_gate\`** + call site at the top of \`run_publish\`. Decision tree: 1. \`opts.rehearsal_registry = None\` → dormant, publish proceeds. Unchanged behavior for every caller who hasn't opted in. 2. \`opts.rehearsal_skip = true\` → proceed with loud warning. No fake-passing receipt is synthesized; auditors see "no rehearsal ran for this plan_id" in events.jsonl. 3. No \`rehearsal.json\` → refuse with fix-suggesting error. 4. Receipt's \`plan_id\` != current plan_id → refuse (stale). 5. \`passed: false\` → refuse with reason. 6. Fresh passing receipt → proceed, log a confirmation info line. - **\`run_rehearsal\` persists the sidecar** on completion. Best-effort: if the sidecar write fails, the events log is still authoritative and the gate will (correctly) refuse, blocking publish — the safe default. ## Tests 13 new tests. All green locally, clippy clean, fmt clean: - 3 roundtrip/persistence tests in \`state::rehearsal\` - 6 gate-helper tests (dormant / skip / missing / stale / failing / fresh-pass) - 1 e2e test that \`run_publish\` actually calls the gate and bails when rehearsal is required but missing Total rehearsal+gate coverage: 17 unit tests. ## Scope — not in this PR - **Install / smoke check** against the rehearsal registry (PR 4 or follow-on). The current gate trusts the rehearsal's publish + visibility verification. - **CI recipe** for kellnr or equivalent sidecar (separate PR; docs work, not code). * docs: how-to for rehearsing against an alternate registry (#97) Stitch the Prove pillar tier 2 together with a step-by-step operator guide that walks through rehearsal-registry configuration, shipper rehearse, the hard gate, and a kellnr-sidecar CI recipe. Covers: - When to use rehearsal (first-publish, post-refactor) - Config via .shipper.toml or CLI flag - Event emissions + rehearsal.json sidecar - Hard-gate decision tree - Complete GitHub Actions example with kellnr sidecar - Troubleshooting for the 3 common gate errors - What rehearsal does NOT cover yet (install/smoke, consumer-build) Tied to docs/README.md so operators can find it. No code changes.

Three-column reconciliation of what's actually merged vs each issue's acceptance checklist. Single source of truth for what closes each pillar issue. **Findings:** - **#90 Recover** — honestly closable. Code side is done (#124 + #130); operator-side real rehearsal is an ops action, not a code gap. - **#97 Prove tier 2** — 85% done. Rehearsal + visibility + hard gate + plan_id binding all landed (#127 + #133). Missing: install/smoke check (cargo install against the rehearsal registry / consumer build). One narrow follow-up PR closes it. - **#98 Remediate** — 60% done. Receipt schema + plan-yank (from-receipt) + yank primitive + fix-forward planning all landed (#121 + #132 + #134). Missing: plan-yank's --starting-crate graph mode, plan execution for yank + fix-forward. Two narrow follow-ups. Also captures the two review concerns on #122 (Trusted Publishing) that were addressed in a follow-up commit to that PR. Recommended next merge order and follow-up PRs spelled out at bottom.

Adds the final piece of #97's acceptance: \`cargo install --registry <rehearsal> <crate>\` after rehearsal publishes succeed, as end-to-end proof that the crate actually resolves and installs via the registry index — the scenario that workspace-path dependencies defeat and that killed the rc.1 first-publish. ## What's new - **EventType variants**: \`RehearsalSmokeCheckStarted\`, \`RehearsalSmokeCheckSucceeded\`, \`RehearsalSmokeCheckFailed\`. - **RuntimeOptions**: \`rehearsal_smoke_install: Option<String>\`. - **CliOverrides + CLI flag**: \`--smoke-install <CRATE>\` (global). Crate must be in the rehearsal plan and have a \`[[bin]]\` target; library-only crates can't be smoke-installed directly (consumer- workspace build variant is a follow-on). - **\`ops::cargo::cargo_install_smoke\`**: wrapper around \`cargo install --registry <name> <crate> --version <v> --root <tmp> --force --locked\`. - **engine::run_rehearsal** post-publish step: if all packages passed AND \`--smoke-install\` names an in-plan crate, install it from the rehearsal registry. Install failure fails the whole rehearsal (sets \`first_failure\` + \`RehearsalComplete { passed: false }\`). Named-but-not-in-plan: warn only; don't fail rehearsal over a typo. ## Tests - \`run_rehearsal_smoke_install_happy_path_emits_succeeded_event\` — flag named an in-plan crate, fake cargo returns 0, assert \`rehearsal_smoke_check_started\` + \`_succeeded\` events. - \`run_rehearsal_smoke_install_missing_target_warns_without_failing\` — flag named a crate not in the plan, assert reporter warn + rehearsal still passes. 19 rehearsal+gate tests pass (up from 17). ## Closes #97 With this PR, #97's issue checklist is 100% fulfilled: - [x] Publish packaged tarballs to alt registry (#127) - [x] Verify presence on rehearsal registry (#127) - [x] Install/build from rehearsal registry (**this PR**) - [x] Record proof in RehearsalComplete event (#127) - [x] Hard gate binding on plan_id (#133) The \`#97\` issue can be closed once this merges.

Three-column reconciliation of what's actually merged vs each issue's acceptance checklist. Single source of truth for what closes each pillar issue. **Findings:** - **#90 Recover** — honestly closable. Code side is done (#124 + #130); operator-side real rehearsal is an ops action, not a code gap. - **#97 Prove tier 2** — 85% done. Rehearsal + visibility + hard gate + plan_id binding all landed (#127 + #133). Missing: install/smoke check (cargo install against the rehearsal registry / consumer build). One narrow follow-up PR closes it. - **#98 Remediate** — 60% done. Receipt schema + plan-yank (from-receipt) + yank primitive + fix-forward planning all landed (#121 + #132 + #134). Missing: plan-yank's --starting-crate graph mode, plan execution for yank + fix-forward. Two narrow follow-ups. Also captures the two review concerns on #122 (Trusted Publishing) that were addressed in a follow-up commit to that PR. Recommended next merge order and follow-up PRs spelled out at bottom.

* feat(rehearse): smoke-install check closes #97 Prove tier 2 (#97 PR 4) Adds the final piece of #97's acceptance: \`cargo install --registry <rehearsal> <crate>\` after rehearsal publishes succeed, as end-to-end proof that the crate actually resolves and installs via the registry index — the scenario that workspace-path dependencies defeat and that killed the rc.1 first-publish. ## What's new - **EventType variants**: \`RehearsalSmokeCheckStarted\`, \`RehearsalSmokeCheckSucceeded\`, \`RehearsalSmokeCheckFailed\`. - **RuntimeOptions**: \`rehearsal_smoke_install: Option<String>\`. - **CliOverrides + CLI flag**: \`--smoke-install <CRATE>\` (global). Crate must be in the rehearsal plan and have a \`[[bin]]\` target; library-only crates can't be smoke-installed directly (consumer- workspace build variant is a follow-on). - **\`ops::cargo::cargo_install_smoke\`**: wrapper around \`cargo install --registry <name> <crate> --version <v> --root <tmp> --force --locked\`. - **engine::run_rehearsal** post-publish step: if all packages passed AND \`--smoke-install\` names an in-plan crate, install it from the rehearsal registry. Install failure fails the whole rehearsal (sets \`first_failure\` + \`RehearsalComplete { passed: false }\`). Named-but-not-in-plan: warn only; don't fail rehearsal over a typo. ## Tests - \`run_rehearsal_smoke_install_happy_path_emits_succeeded_event\` — flag named an in-plan crate, fake cargo returns 0, assert \`rehearsal_smoke_check_started\` + \`_succeeded\` events. - \`run_rehearsal_smoke_install_missing_target_warns_without_failing\` — flag named a crate not in the plan, assert reporter warn + rehearsal still passes. 19 rehearsal+gate tests pass (up from 17). ## Closes #97 With this PR, #97's issue checklist is 100% fulfilled: - [x] Publish packaged tarballs to alt registry (#127) - [x] Verify presence on rehearsal registry (#127) - [x] Install/build from rehearsal registry (**this PR**) - [x] Record proof in RehearsalComplete event (#127) - [x] Hard gate binding on plan_id (#133) The \`#97\` issue can be closed once this merges. * test: rebaseline shipper-config snapshots for rehearsal_smoke_install field * test: rebaseline shipper-types RuntimeOptions debug snapshot for smoke-install field

gemini-code-assist Bot reviewed Apr 17, 2026

View reviewed changes

chatgpt-codex-connector Bot reviewed Apr 17, 2026

View reviewed changes

EffortlessSteven mentioned this pull request Apr 17, 2026

feat(rehearse): hard gate — run_publish requires passing rehearsal (#97 PR 3) #128

Closed

4 tasks

EffortlessSteven force-pushed the feat/97-rehearse-execution branch from 827ca17 to 87c5604 Compare April 18, 2026 00:09

EffortlessSteven merged commit a84209f into main Apr 18, 2026
19 checks passed

EffortlessSteven deleted the feat/97-rehearse-execution branch April 18, 2026 00:11

EffortlessSteven mentioned this pull request Apr 18, 2026

feat(rehearse): hard gate — run_publish requires passing rehearsal (#97 PR 3) #133

Merged

EffortlessSteven mentioned this pull request Apr 18, 2026

feat(rehearse): smoke-install closes #97 Prove tier 2 (#97 PR 4) #137

Merged

4 tasks

EffortlessSteven mentioned this pull request Apr 18, 2026

Rehearsal registry: strengthen preflight from dry-run to actual publish + install proof #97

Closed

10 tasks

This was referenced Apr 19, 2026

Competency: Prove — gap audit and roadmap #100

Open

Roadmap: post-v0.3.0-rc.1 product scorecard across nine competencies #109

Open

EffortlessSteven mentioned this pull request Jun 24, 2026

ci(deps): bump codecov/codecov-action from 6 to 7 #404

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

feat(rehearse): engine::run_rehearsal + shipper rehearse CLI (#97 PR 2)#127

feat(rehearse): engine::run_rehearsal + shipper rehearse CLI (#97 PR 2)#127
EffortlessSteven merged 1 commit into
mainfrom
feat/97-rehearse-execution

EffortlessSteven commented Apr 17, 2026

Uh oh!

coderabbitai Bot commented Apr 17, 2026 •

edited

Loading

Rate limit exceeded

Uh oh!

gemini-code-assist Bot left a comment

Uh oh!

gemini-code-assist Bot Apr 17, 2026

Uh oh!

gemini-code-assist Bot Apr 17, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

chatgpt-codex-connector Bot Apr 17, 2026

Uh oh!

chatgpt-codex-connector Bot Apr 17, 2026

Uh oh!

codecov Bot commented Apr 17, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Conversation

EffortlessSteven commented Apr 17, 2026

Summary

Key invariants

Scope — explicitly NOT in this PR

Tests

Test plan

Uh oh!

coderabbitai Bot commented Apr 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Rate limit exceeded

Uh oh!

gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

gemini-code-assist Bot Apr 17, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot Apr 17, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector Bot Apr 17, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector Bot Apr 17, 2026

Choose a reason for hiding this comment

Uh oh!

codecov Bot commented Apr 17, 2026

Codecov Report

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

coderabbitai Bot commented Apr 17, 2026 •

edited

Loading