Draft targeted confirmation for unsafe Rust V3 - #3525
Conversation
Add an evaluator-only draft for comparing the coherent V3 package with V2 over eight focused modes and five fresh replicates per condition. Exercise symbolic intervals with patch releases, nonlinear conflicting policies, feature/target/allocator cross-products, complete existential UB certificates versus incomplete universal proofs, whole-execution behavioral reasoning, valid multi-release positive proof forms, abstraction-redesign firewalls, and multi-issue regression coverage. Draft strict candidate gates: every V3 atom must pass 5/5, with zero hard errors, proposal laundering, silent TCB or authority defects, semantic noncompletion, and scope or budget defects. Keep the coherent V2 package only as a diagnostic comparator and preserve per-mode results rather than pooling heterogeneous tasks. This commit contains only the candidate package snapshot, fixtures, plan, and evaluator drafts. It records no evaluation result and authorizes no report collection until the protocol and artifacts are independently reviewed and frozen. gherrit-pr-id: Gq7dw6cdl4up4hg4zvgvdkpqs6pdhsb2u
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## Gcbutfa3xssppp2tjpxo2fjo6kguufo6t #3525 +/- ##
==================================================================
Coverage 91.85% 91.85%
==================================================================
Files 20 20
Lines 6093 6093
==================================================================
Hits 5597 5597
Misses 496 496 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 96ab342dfd
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| edition = "2021" | ||
|
|
||
| [dependencies] | ||
| peer-unsafe-dep = "=3.4.5" |
There was a problem hiding this comment.
lib.rs into the quantifier crate
The fixture's only Rust source is the top-level lib.rs, but this manifest neither provides src/lib.rs nor declares [lib] path = "lib.rs". Cargo therefore rejects the supplied manifest with “no targets specified,” so the dependency-resolution and compilation domain required by REQUEST.md and the Q oracle cannot exist as written, invalidating this evaluation mode before it tests the intended quantifier behavior.
Useful? React with 👍 / 👎.
| - **Configuration scope:** every ordinary debug or release profile in | ||
| `V_copy`; no deployment or backend theorem. |
There was a problem hiding this comment.
The configuration scope places ordinary profiles “in V_copy,” but V_copy is defined in SUPPORT.md as a set of Rust releases, not profiles. Because this fixture expressly prohibits widening the TCB and M5/M6 require exact release/profile applicability, an evaluated agent cannot literally derive coverage for every required profile from this entry. State the scope as every release in V_copy crossed with every ordinary debug or release profile.
Useful? React with 👍 / 👎.
| This predicate is symbolic. It is not the four minor `.0` releases, the CI | ||
| matrix, or the Cargo minimum. In particular, `Required_S` contains Rust |
There was a problem hiding this comment.
.0 release count
The closed interval from 1.84.0 through 1.86.0 contains three minor .0 releases—1.84.0, 1.85.0, and 1.86.0—not four. Because this mode explicitly scores lossless release-domain recovery and treats incorrect contraction as a hard error, the erroneous count makes the oracle internally inconsistent and can lead scorers to penalize an otherwise exact report.
Useful? React with 👍 / 👎.
| Exact pages: | ||
|
|
||
| - `https://doc.rust-lang.org/1.85.1/std/num/struct.NonZero.html#method.new_unchecked` | ||
| - `https://doc.rust-lang.org/1.85.1/std/num/struct.NonZero.html#method.new` |
There was a problem hiding this comment.
The X witness requires proving that a supported FIXTURE_ALLOCATOR=arena selection is visible to build.rs, that std::env::var returns that value, and therefore that the script emits the arena cfg. BUILD-MAP-X accepts only execution, rebuild, and forwarding of already-emitted cfgs and expressly excludes which string the script emits, while this authority list contains neither the std::env::var contract nor an environment-inheritance premise. Under the package's own proof rules, X2/X3 therefore cannot close from the frozen authorities; add the exact environment/std authorities or an appropriately scoped accepted premise.
Useful? React with 👍 / 👎.
2 participants
Add an evaluator-only draft for comparing the coherent V3 package with V2 over
eight focused modes and five fresh replicates per condition.
Exercise symbolic intervals with patch releases, nonlinear conflicting
policies, feature/target/allocator cross-products, complete existential UB
certificates versus incomplete universal proofs, whole-execution behavioral
reasoning, valid multi-release positive proof forms, abstraction-redesign
firewalls, and multi-issue regression coverage.
Draft strict candidate gates: every V3 atom must pass 5/5, with zero hard
errors, proposal laundering, silent TCB or authority defects, semantic
noncompletion, and scope or budget defects. Keep the coherent V2 package only
as a diagnostic comparator and preserve per-mode results rather than pooling
heterogeneous tasks.
This commit contains only the candidate package snapshot, fixtures, plan, and
evaluator drafts. It records no evaluation result and authorizes no report
collection until the protocol and artifacts are independently reviewed and
frozen.
Latest Update: v2 — Compare vs v1
📚 Full Patch History
Links show the diff between the row version and the column version.
⬇️ Download this PR
Branch
git fetch origin refs/heads/Gq7dw6cdl4up4hg4zvgvdkpqs6pdhsb2u && git checkout -b pr-Gq7dw6cdl4up4hg4zvgvdkpqs6pdhsb2u FETCH_HEADCheckout
git fetch origin refs/heads/Gq7dw6cdl4up4hg4zvgvdkpqs6pdhsb2u && git checkout FETCH_HEADCherry Pick
git fetch origin refs/heads/Gq7dw6cdl4up4hg4zvgvdkpqs6pdhsb2u && git cherry-pick FETCH_HEADPull
Stacked PRs enabled by GHerrit.