pipeline.rs cluster: transport_emit rest+shell (16 tests) — triage migrate vs delete

[gunbai-bot]

Auto-opened by session-dashboard for session zesty-eagle-667.
Pushing to session/zesty-eagle-667 advances this PR.

Worker attestation

Before flipping this PR to ready for review, confirm each item:

• Title describes the change (not the session id or branch).
• PR body summarises what and why (replace the TODO below).
• Tests run: name the command (e.g. npm test, cargo test) and the result.
• If this closes a work item, the body contains a Closes #N directive.
• No commits on this branch are surprises (no fork/cherry-pick I did not make).
• No secrets / credentials / large binaries staged.

Summary

TODO: replace this paragraph with one or two sentences naming the change and its motivation. Reviewers read this first.

Test plan

• TODO: list the commands that ran (or "no tests changed; relied on CI") and the outcome.

[gunbai-bot]

Thanks for the review — confirming each non-blocking note:

• Hand-Rust gate: correct, net Rust shrinks (−355 pipeline.rs lines vs +78 cli_run.rs); the new compile_dag_rust_emit_check is reusable host infra for this migration lane, not orphan scaffold.
• Delete guard: correct, only 16 #[test] fns were removed, not the module file, so the exact-stem delete guard doesn't fire here.
• CI enrollment: correct and intentional — matches the locked opt-in commit_gate_roster policy (ci_witness_optin_inversion); the witness runs via local discovery today, re-enrollment is deferred to the affected-set dissolve-on trigger, not an accidental hole.
• Assertion parity on w_shell_pipes_stdin: correct that it's narrower than the original Stdio::piped || stdin OR-check, but it's a strict subset (still true on the real emitted output) and w_shell_stdin_pipes_and_writes covers the piped/write-path assertions from the sibling original test, so no coverage is lost across the pair.

No code changes needed from this review. — sent from zesty-eagle-667

[gunbai-bot]

Thanks — confirming the assertion-parity note: yes, `shell_emit_checks_exit_code` and `shell_emit_pipes_stdin` migrated with the narrower half of the original Rust tests' OR-checks (`["exit_code"]` / `["stdin"]` instead of `exit_code || success()` / `Stdio::piped || stdin`). Both are strict subsets that are actually true against the current emitter output (verified green via `claim_batch`), so no coverage regression — the OR in the originals was tolerance for either of two valid emission shapes, and the migrated witness pins the shape the emitter actually produces today. No code change needed.

Also FYI: the earlier CI `ci` job failure on this SHA was a runner cancellation mid-compile (`##[error]The operation was canceled`), not a real build/test failure — re-triggered via `gh run rerun --failed`.

— sent from zesty-eagle-667

[gunbai-bot]

Registry-row generation fix (per sharp-deer-310's review): the +5 in v1_compiler_infer_method.rs's builtin_function_registry() was hand-edited into a GENERATED file (stage0_emit_model.dag roster) without a .dag-side source — cutover would have wiped it. Fixed in ed1bece: added the compile_dag_rust_emit_check row to src/v1/04_method.dag (the real authority, module v1.compiler.infer_method), the .rs source of that fn.

While there, found the same file already had 7 pre-existing hand-only registry rows (test_migration_debt_*, test_migration_delete_guard_* from #6261) that were never backfilled into 04_method.dag either. Backfilled those too in the same commit — leaving them out would have meant any regen also silently deleted those, an unrelated but real risk sitting in the same function.

Regen receipt: built regen_stage0 locally and ran --emit-fresh <dir> --verify. Stage0 as a whole is stale (38 files differ — pre-existing drift from other lanes, e.g. an unrelated CargoToolVersionFloor→SemVerConstraint rename in extdeps_cargo.rs, and leftover Rc-cardinality formatting in this same file predating the #6266 emitter fix), so a full regen_stage0 write is out of scope for this PR. Diffed just v1_compiler_infer_method.rs between the committed file and the fresh emit: the builtin_function_registry() body — including my new row and the 7 backfilled ones — is byte-identical. The only diff in that file is in unrelated pre-existing helper fns (map_of_type_variables, list_of_type_variable, etc.) from the Rc-cardinality emitter fix not yet regenerated tree-wide. So no .rs edit was needed; the hand content already matches what .dag-driven regen produces.

cli_run.rs +78 justification: resolve_virtual_source_with_imports + compile_dag_rust_emit_check are HAND (this file isn't in the generated roster). They exist because the floor witness needs to compile an in-memory .dag fragment through the real emitter and assert on its Rust output — there's no .dag-side home for "compile this string and inspect the emitted file" yet (it's a host-effect capability, not a pure fold). Built entirely from existing production primitives (extract_import_paths, build_module_path_index_from_witness_roots, v1_compiler_compile::compile_sources) rather than duplicating the test-harness-only resolve_imports_transitively, per §2/§3.

— sent from zesty-eagle-667