10 lines
1.7 KiB
Markdown
10 lines
1.7 KiB
Markdown
# Allocation and repair baseline — Windows, 2026-09-12
|
|
|
|
Source **d9b3f0dc91e8882e6efc262c2e8d971998d2694e** passes **792/792** desktop tests, with no failed or ignored cases. The five saved-destination repair regressions pass in the combined App: invalid Open and rollback bind dirty state to the actual destination while retaining model/history; successful Save records the new baseline. Allocation planner, pointer dispatch and four actual App-tree tests also pass their recorded assertions.
|
|
|
|
Command: `cargo test --locked -p oa-desktop --bin arcadia-studio -- --test-threads=1`, using the existing Windows GNU LLVM toolchain (`.local/toolchain/env.ps1`, Rust/Cargo 1.98.1, LLVM-MinGW 20260908). Run began 11:18:56 UTC and ended 11:26:43 UTC; tests took 296.11 seconds. The runner verified all **526** source inputs byte-for-byte against Git before execution and their SHA-256 identities afterward; none changed. The input manifest SHA-256 is `5183c5000c41bd392733b111131e3b2f6da084bc58e92002cdf5b14e29512f77`.
|
|
|
|
[Exact log, receipt, input manifest and source review](experiments/allocation-d9b3f0d-windows-20260912/archive-manifest.json) are preserved without rewriting their bytes. The source review establishes an additional collapsed-target reachability defect that the passing suite does not cover. The [feedback decision](release-allocation-feedback-decision-20260912.md) rejects that implementation's fixed-label refusal and defines the pending correction. A passing baseline is not acceptance of that known defect.
|
|
|
|
No strict Clippy, matching GUI build, native or packaged allocation acceptance is claimed for this source. Linux/WSL remains user-paused. The report does not transfer evidence to later source changes or close the full OA/SA objective.
|