Repository navigation
XL-0-SERVICE: the service-emission path in 05_emit_rust binds stdout to every declared output field and does not box the error arm — it emits non-compiling Rust - #9887
Conversation
…h a three-output known-hole probe The shell service-emission path in 05_emit_rust binds `stdout` to EVERY declared output field, whatever source the declaration named, and returns a bare String from the error arm against a declared Box<dyn std::error::Error>. Both emit Rust that does not compile, with zero diagnostics. This commit lands the reproduction only -- the lane was parked by the operator before the repair (it does not reduce the typeck count). The probe is a §4b(4) known-hole probe: green today asserting the WRONG behavior, and it must flip to the refusal assertion when the wall lands. The three-output fixture is the load-bearing part. Two fields cannot distinguish "wrong tuple index" from "the declared source never arrived"; three can, and the ABSENCE of the emitted `let stderr = ...` prelude is the positive evidence that the `from "stderr"` field was invisible rather than mis-ordered. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SKidppJZFSPywEdbVGJ1Ex
…egenerate the stage0 mirror
The parked commit hand-wrote the probe into src/v1/stage0/src/compiler_tests.rs,
which is a GENERATED file -- the emitted-population manifest names it, so the next
regeneration would have silently deleted it. The authority for that surface is
src/v1/compiler_tests_rust.dag.
This commit authors ct_shell_service_output_projection_known_hole_probe_test()
there, enrolls it in compiler_tests_source(), and carries both generated
projections that follow from it:
- src/v1/stage0/src/v1_compiler_compiler_tests_rust.rs (the mirror of the
generator module itself), and
- src/v1/stage0/src/compiler_tests.rs (the generator's own output), where the
probe now sits at the position compiler_tests_source() places it rather than
at end-of-file.
Both were taken from --required-regen candidate bytes, not hand-edited: round one
reported exactly one generated-surface drift (v1_compiler_compiler_tests_rust.rs),
and round two -- with the seed rebuilt from that mirror -- reported exactly one
more (compiler_tests.rs). The probe's own bytes are unchanged; it still asserts
today's WRONG behavior and must flip when the wall lands.
…e-emission fabrication -- kept for reference only, not for landing
|
Closing this — it is a duplicate dispatch, not a competing proposal. sharp-dove-805 owns the repair for this defect and is carrying it in #9886; I own the acceptance The branch is not deleted, so the commit stays reachable. It contains an independently built My acceptance verdict on #9886 is recorded with the manager: arm B (the fabricating channel arm) is — sent from calm-boar-314 |
Auto-opened by session-dashboard for session
calm-boar-314.Pushing to
session/calm-boar-314advances this PR.Worker attestation
Before flipping this PR to ready for review, confirm each item:
npm test,cargo test) and the result.Closes #Ndirective.Summary
TODO: replace this paragraph with one or two sentences naming the change and its motivation. Reviewers read this first.
Test plan