Repository navigation
row 155 F25 REHOME: misc non-money rows (plane door, egress ports, auth, metrics) - #910
Merged
Merged
Conversation
Ports coverage-map-v2 REHOME rows to the served plane door and the composition root, each test under its legacy name with the figures the map records and a RED line: plane-llm: alloc_gate (39, 40), exchange_shaping (43), plane_door (98, 99, 100, 375, 643), exchange_probe (220, 230), exchange_user_agent (325), refusal_envelopes (330, 682). root: egress_ports (44, 45, 46, 47, 542, 612), declared_credentials (73), linked_auth (402), door_steps (463), policy (603), docs_examples (101), metrics_scrape_boot_window (137), metrics_routes_across_export_reload (138), metrics_cross_plane (604). Tests only; not run locally.
… proves; the reopen on a hot apply is the egress lane's (coordinator ruling)
This was referenced Oct 14, 2026
Collaborator
Author
|
#1028 also edits crates/busbar-plane-llm/tests/exchange_probe.rs, exchange_user_agent.rs and refusal_envelopes.rs (a local TestHost/HOST, and |
Collaborator
Author
|
Rule 20 note from #1031 (row 155, draft): disjoint hunks in tests/exchange_shaping.rs and root/tests/door_steps.rs (one test each). A test merge is clean. |
…e root names no auth style PB-4, PB-8, PB-66: four rehomed tests shared their legacy twin's name, so the design bindings' bare test refs named no particular test. Renamed: - egress_ports the_queue_frees_one_permit_to_exactly_one_of_two_waiters - egress_ports a_members_attempt_timeout_wins_over_its_models_default - exchange_user_agent every_dialects_egress_user_agent_version_is_pinned_and_present - metrics_cross_plane the_scrapes_pool_queued_gauge_reads_the_live_depth The bindings keep citing the legacy tests until C repoints them. law0 busbar x auth / instance:auth: row 73's keyless test no longer names an auth style. It binds the fixture's non-signing bindings (a signing binding is the one that sends fields of its own) and covers the token override through the dialect default, which is what the door binds for it. Row 402's test hands the request-path module its credential beside an empty head instead of building a Bearer line. Merged origin/predev.
…d from the fixture (no coverage lost to law0) An operator's provider `auth:` override is live on the door: root door_steps.rs:1590-1597 (style_word), :1622 (ProviderRoute.style) and :2458-2468 (member_routes binds that exact style, not the dialect default). The keyless test binds the token override again with its style read from testing/plane-copies/declared-credentials.json `override_styles.token`, so the root names no auth style. The full assertions are restored: three or more bindings; no header for an empty key in both modes; one header for a real credential, on the override too. The signing exclusion is binding.sent.is_empty(), which excludes exactly the one signing binding the fixture records.
…ls an auth plugin's crate id (law0 busbar x auth 51 -> 49) The rehomed test kept its legacy name test_keyless_lane_sends_no_auth_header, and the name and its doc citation each count one busbar x auth hit (an auth plugin's crate id). Renamed to a_keyless_lane_presents_no_credential_field; the doc cites the legacy test by its file and map row.
…busbar_contract::transport (as root/tests/door_steps.rs names it); merged origin/predev
…e before they parse a models/pools deployment; the scrape cells assert a series a fresh node renders
egress_ports sealed_timeouts parsed a deployment naming models/pools with no plane registered
("no plane is registered to own this config section"); it registers the kernel's neutral test
plane first (test_support::register_neutral_test_plane), as kernel config tests do.
Rows 137/138: the legacy tests seeded busbar_requests_total in process; the rehomed tests spawn a
node with no traffic, so they assert the scrape-time lane series busbar_lane_state the exposition
always carries. ledger_identity's oracle panic is the removed hand-installed oracle (environment).
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
1 participant
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.
Row 155, F25 REHOME (coverage map v2). Ports each REHOME row below to its 1.6 home at equal strength: the legacy name, the map's figures, and a RED line in the doc comment. The changes are tests only.
Rows (map row -> destination)
Not run locally. Per queue rules there was no cargo build or test on the laptop. ModernCI is the proof.
Neutrality. The root files name no dialect, vendor or plane instance. Legacy references read "the retiring engine crate's
src/...". Dialect and protocol names are read as data: the node fixture, the binary's ownmust be one of:refusal, and the registry's default dialect. The boot-window scrape reuses its existing connection helper, so no new transport word is added.RED-able tests (suspected live divergence)
a_changed_upstream_timeout_rebuilds_the_client_an_unrelated_apply_reuses_itis held on the settings a transport door is opened with. The served connector is built once at boot (root/connector.rsboot/install), so a hot-appliedlimits.upstream_request_timeout_secsnever reopens the transport doors. That is a suspected 1.5.5 divergence, and this test cannot see it.test_keyless_lane_sends_no_auth_header: red if a linked auth plugin presents a field (for example a bareBearer) for an empty credential. The door path has no writer-side empty-key guard.test_metrics_requires_auth_in_chain_mode: red if/auth/token?logout=1is not served open on the data listener.Rows not in this PR. 78, 743, 744, 746 and the on_driver rows need the llm door linked in the root, so they go on a lane-p3-llm-default-based branch. 560-562 (golden_parity against the 1.5.5 recording) are not ported; see the lane report.
Row 603 (coordinator ruling 2026-10-14 ~02:55Z): this PR pins only the settings a transport door is opened with. The test is renamed
a_changed_upstream_timeout_changes_the_door_settings_an_unrelated_apply_does_not. A hot-appliedlimits.upstream_request_timeout_secsreaching the transport doors (1.5.5 hot-reload) belongs to the egress lane, which owns root/connector.rs and the RED for it.Row 73 (coordinator, 2026-10-14): the keyless test covers the operator's
auth:override too, because the door binds that exact style (root door_steps.rs:1590-1597style_word, :1622ProviderRoute.style, :2458-2468member_routes). The override's style word is read fromtesting/plane-copies/declared-credentials.jsonoverride_styles.token, so the root names no auth style.