Repository navigation
test: avoid unused BS-07 mutant preparation on other hosts - #448
Conversation
|
Release qualification is now consolidated in draft #450 at 25be1d8. It retains this PR's reviewed commit as a parent, carries the Atlas and BS-07 changes together with the Intel construction/partition fix, and targets release/v0.4. Keep this PR open and its history intact until the exact combined candidate passes full acceptance and is merged; do not launch another independent full gate here. #450 currently has source review and focused local tests only, with real Rust/Intel qualification still pending. |
|
The combined candidate #450 has now landed in release/v0.4 as cf76090 after all 50 exact full-gate jobs and exact-head reproducibility passed. Fresh readback confirms actual tree 0c24b969dcc5c02d9afb7616f97736827b7b09e6 equals the reviewed and fully tested tree. This preserves both source histories; GitHub now reports this original carrier merged. No separate rerun is needed. Acceptance: #450 (review) |
Closes #447.
Change
Prepare BS-07's 150 evaluator mutants and 1,500 rustc mutants only after the existing normative-host return. Non-normative hosts never execute those corpora; previously they still performed oracle lowering and outcome serialization for every accepted mutant.
The all-host contract is unchanged: eligibility, rule validity/dead-function evidence, exact 22 rule / 350 vector / 292 evaluator-size counts, and all four native differential roots still execute. Linux x86-64 retains the exact seeded generators and the complete evaluator, native-mutant, twenty-plant, export, and committed-evidence checks. No timeout, test inventory, count, seed, language identity, or workflow changes.
Focused verification
Qualification limits: No Rust toolchain is available in the editing executor. Rust compilation, focused BS-07 execution, and final-head Linux/Intel acceptance are pending. Draft CI is development evidence, not the full acceptance gate. No runtime saving is claimed or measured; this removes proven unused work but does not establish that it resolves the four-hour Intel timeout in #444.
Release and ownership
Targets the active release/v0.4 line. The eventual qualified release landing must be carried forward through the existing #405 integration lane.
This overlaps the BS-07 count assertion hunk owned by #442. It deliberately separates the base-vector counts from mutant counts but does not implement or change the count-writer design. When integrating #442, preserve its base-count reporting before the host return and keep the mutant preparation/assertion after it.
The broader one-construction / disjoint host-partition design remains separate. Avoid another full monolithic rerun while that release qualification plan is being prepared.