| FazBrowse GitHub Viewer | Trending | | Home |
| Tools: [Download Repo ZIP] [Original HTTPS Page] |
Sorry, something went wrong.
|
Follow-up note from closing #1014: PR #1014 attempted to fix array→wrapper pointer cast projections ([u8; 2] → Wrapper([u8; 2]), nested wrappers, singleton-wrapped arrays) via #pointeeProjection / #wholeArrayTargetCompatible changes, but those rules were never reached — proofs stuck at thunk(#cast(...)) identically with or without the changes. This PR (#1003) resolves the simple wrapper case: ptr-cast-array-to-wrapper-fail is promoted to a passing test. However, two cases still carry the -fail suffix with modified (but not fully passing) expected output:
These two may need a follow-up to handle recursive wrapper/array projection in the target type during pointer casts. |
Sorry, something went wrong.
| Back | FazBrowse Home | New Git URL |
Summary
Refactors the pointer-to-integer work in #1002 into a reviewable red/green history and fixes the remaining symbolic regressions in the address model.
This PR now does four things:
Implementation highlights:
Commit Guide
Verification
Local validation run on the rewritten branch:
uv --directory kmir run pytest src/tests/integration/test_integration.py -v --timeout=600 -k 'alignment-check or interior-mut3 or local-raw or ptr-through-wrapper or ptr-transmute-nonzero or ptr-transmute-two-locals or ptr_offset or raw-ptr-cast or ref-ptr-cast-elem-offset or ref-ptr-cast-elem or ptr-cast-cross-frame-eq'Result:
Additional proof checks used while validating the migrated regressions:
GitHub Actions on the rewritten history are currently in progress and should be treated as the remaining readiness gate.
Related