Skip to content

llm codec tests: F25 REHOME rows 108, 110, 350, 504, 505, 613, 614 - #883

Merged
MattJackson merged 1 commit into
predevfrom
lane-llm-codec-rehome
Oct 15, 2026
Merged

MattJackson merged 1 commit into
predevfrom
lane-llm-codec-rehome

Conversation

@MattJackson

Copy link
Copy Markdown
Collaborator

What. These F25 coverage-map v2 REHOME rows have their home under busbar-plane-llm/src/codec/**. Each legacy crates/busbar-llm/src/engine/tests/* test is ported at equal strength: the same inputs, the same literal figures, and no circular check. The assertions run on the served codec path the row names.

row legacy test ported to
108 egress_dropped_controls_audit_tests.rs:89 bedrock/tests/tests.rs::openai_tool_choice_none_onto_bedrock_forwards_and_reports_the_drop: non-empty body, get_weather kept, toolChoice omitted, tool_choice=none in dropped controls
110 egress_dropped_controls_audit_tests.rs:245 anthropic/tests/ir_mapping_q57_tests.rs::native_structured_output_lane_drops_schema_less_json_and_reports_it
350 ingress_indistinguishability_tests.rs:2234 tests/proto/stream_translate_tests.rs::an_anthropic_tool_call_id_reaches_an_openai_client_remapped_and_unleaked: call_…, not toolu_origRAW123, raw id absent from the body, decodes back
504 mid_stream_error_tests.rs:294 tests/wire_shim_tests.rs::strip_router_shim_keys_removes_the_never_native_keys_per_egress
505 mid_stream_error_tests.rs:337 tests/wire_shim_tests.rs::shim_strip_then_rewrite_model_keeps_the_model_cross_protocol (plus the same-dialect relay byte check)
613 stop_sequence_cap_degrade_tests.rs:23 tests/proto/response_format_matrix_tests.rs::openai_to_cohere_over_cap_stop_sequences_are_clamped_not_rejected: six become [a..e]
614 stop_sequence_cap_degrade_tests.rs:63 …::openai_to_cohere_exactly_cap_stop_sequences_are_forwarded_whole

For rows 108 and 110, the legacy audit row (egress.control_unrepresentable) is fed by the dropped-control name, so the codec-level port pins that name. That is the "or dropped_egress_controls contains it" the row allows.

Rows 107 and 109 are in a separate PR stacked on #857, because #857 restores 1.5.5's response_format drop on Anthropic. Tests only; not $.

…rted to their codec homes

Each legacy busbar-llm engine test is ported at equal strength (its literal figures) onto the
served codec path its F25 v2 row names: Bedrock tool_choice none forwards and reports the drop
(108); a native-structured-output Anthropic lane rides output_config.format and reports a
schema-less json_object (110); an Anthropic tool id reaches an OpenAI client remapped and unleaked
(350); the router shim keys strip per egress and the model survives strip-then-rewrite, with the
same-dialect relay sending the caller bytes (504, 505); the Cohere stop-sequence cap clamps six to
five and forwards five whole (613, 614).
@MattJackson

Copy link
Copy Markdown
Collaborator Author

Rule 20 note from #1031 (row 155, draft): one disjoint hunk in stream_translate_tests.rs (synthesize_from_response takes the cap). A test merge is clean.

@MattJackson
MattJackson merged commit efd1832 into predev Oct 15, 2026
1 check was pending
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant