…e-note
Final pass over CRANELIFT_IMPLEMENTATION.md to remove wording that reads as open
work but is done, and to resolve the one adjacent opportunity the doc flagged:
- Step 8 title "(DONE, drain pending)" -> "(DONE)" (the `abi::Intrinsic` enum is
fully deleted; grep confirms no `Intrinsic`/`Member.intrinsic` remains).
- 10-rc "The fix (commit pending)" -> "(shipped, PR #1972)".
- Slice-2 "Still to validate: GC stack-map coverage for byte-defined frames" ->
resolved: the AOT-L3 investigation confirmed there are NO precise stack maps
(`stack_map_registry` dead); JIT and AOT share one conservative scanner covering
both frame kinds, suite passes — a non-issue, not a pending gate.
- `Atomics` primordial: was "blocked on step 5" (now done). DEFERRED with
justification — not a numbered step, niche, and by Step 4's measured lesson the
atomic instruction is ~1 ns so the call-removal win is modest; revisit only if a
real workload shows Atomics hot.
The deep slice-2 build-saga narrative (a historical journey log closed by "SLICE 2
COMPLETE") is left as the record. Doc-only.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Claude-Session: https://claude.ai/code/session_019WP5GhgeEqjR9VmiGst2Pw
Final consistency pass: remove wording that reads as open but is done (Step 8 drain, 10-rc commit-pending, slice-2 GC validation) and resolve the one adjacent opportunity the doc flagged — the
Atomicsprimordial inline (unblocked by step 5, but DEFERRED with justification: not a numbered step, niche, ~1 ns instruction per Step 4's lesson). Doc-only.🤖 Generated with Claude Code