docs: mark W46 Phase 1 done and refresh memo

.dev/checklist.md

@@ -10,23 +10,6 @@ Prefix: W## (to distinguish from CW's F## items).
 
 ## Open Items
 
-- [ ] W46: Un-link libc — migrate WASI I/O / cache / platform off `std.c.*` onto
-  `std.Io` (or `std.posix.system` direct syscalls). Scope: `src/wasi.zig`
-  (fd_read / fd_write / fd_pread / fd_pwrite / fd_seek / fd_tell / fstatat),
-  `src/cache.zig` (fsync), `src/platform.zig` (statx helpers), test helpers.
-  Why this is a priority, not cosmetic:
-  1. **Zig 0.16 philosophy** — release notes explicitly say "go higher
-     (std.Io) or go lower (std.posix.system)"; `std.c.*` is neither direction.
-  2. **Windows correctness** — `std.c.fd_t` = `windows.HANDLE` (pointer),
-     but `std.c.write` binds to MSVCRT `_write(int fd, …)`. The ABI
-     mismatch is what broke Windows real-world compat in v1.10.0.
-  3. **Binary size** — Linux ELF pays ~290 KB for libc linkage; returning
-     to no-libc lets us restore the 1.50 MB guard.
-  4. **Performance** — 2025-02 Andrew Kelley devlog: "No-Libc Zig Now
-     Outperforms Glibc Zig" — no-libc is the fast path.
-  Non-goal for v1.10.x: rewriting all I/O atomically. Plan is phase-by-phase
-  (phase 1 = WASI stdio Windows fix, phase 2 = fd_read/fd_write, …).
-
 - [ ] W45: SIMD loop persistence — Skip Q-cache eviction at loop headers.
   Requires back-edge detection in scanBranchTargets.
 
@@ -54,5 +37,14 @@ W43: SIMD v128 base addr cache (SIMD_BASE_REG x17). Phase A of D132.
 W44: SIMD register class — Q16-Q31 (ARM64) + XMM6-XMM15 (x86) cache.
      Phase B of D132. Merged 2026-03-26. Q-cache with LRU eviction + lazy
      writeback. Benefit limited by loop-header eviction (diagnosed same day).
+W46: Un-link libc — Phase 1 complete (delib 1a–1f, merged 2026-04-24/25).
+     link_libc=false across lib / cli / tests / examples / e2e / bench / fuzz;
+     C API targets (shared-lib, static-lib, c-test) keep link_libc=true because
+     `src/c_api.zig` uses `std.heap.c_allocator`. Platform helpers added:
+     pfdWrite, pfdRead, pfdClose, pfdDup, pfdDup2, pfdPipe, pfdSleepNs, pfdErrno,
+     pfdFsync, pfdReadlinkAt (Linux=direct syscalls, Mac=libSystem auto-link,
+     Windows=kernel32/Win32). Payoff measured: Mac 1.38 MB, Linux 1.65 MB
+     stripped (vs 1.80 MB ceiling). The 1.50 MB target is deferred to the
+     std.Io migration (tracked separately if it resurfaces).
 
 W2-W36: See git history. All resolved through Stages 0-47 and Phases 1-19.

.dev/memo.md

@@ -4,44 +4,46 @@ Session handover document. Read at session start.
 
 ## Current State
 
-- **Zig toolchain**: 0.16.0 (migrated from 0.15.2, 2026-04-24).
-- Stages 0-46 + Phase 1, 3, 5, 8, 10, 11, 13, 15, 19, **20** complete.
+- **Zig toolchain**: 0.16.0 (migrated 2026-04-24).
+- Stages 0-46 + Phase 1, 3, 5, 8, 10, 11, 13, 15, 19, 20 complete.
 - Spec: 62,263/62,263 Mac+Ubuntu (100.0%, 0 skip).
 - E2E: 796/796 Mac+Ubuntu, 0 fail.
-- Real-world: Mac 50/50, Ubuntu 50/50, 0 crash.
+- Real-world: Mac 50/50, Ubuntu 50/50, Windows 46/46, 0 crash.
 - FFI: 80/80 Mac+Ubuntu.
 - JIT: Register IR + ARM64/x86_64 + SIMD (NEON 253/256, SSE 244/256).
 - HOT_THRESHOLD=3 (lowered from 10 in W38).
-- Binary: 1.29MB stripped. Memory: ~3.5MB RSS.
+- Binary stripped: Mac 1.38 MB, Linux 1.65 MB (ceiling 1.80 MB). Memory: ~3.5 MB RSS.
 - Platforms: macOS ARM64, Linux x86_64/ARM64, Windows x86_64.
-- **main = stable** (currently v1.9.1). v1.10.0 on `develop/zig-0.16.0`, awaiting PR.
+- **main = stable**. v1.10.0 released; post-release work on delib / W46 merged
+  via PRs #47 (1a/1b pre-cursor), #48 (1b), #49 (1c/1d/1e/1f + C-API libc fix).
+- link_libc = false across lib / cli / tests / examples / e2e / bench / fuzz.
+  C-API targets (shared-lib, static-lib, c-test) keep link_libc = true because
+  `src/c_api.zig` uses `std.heap.c_allocator`.
 
 ## Current Task
 
-**v1.10.0: Zig 0.16.0 migration — DONE on both platforms (2026-04-24)**
+**W46 Phase 1 — DONE (2026-04-25).** PR #49 merged (`30a3680`). See
+`.dev/checklist.md` Resolved section for details. Next candidate work:
 
-Full rundown in `@./.dev/archive/zig-0.16-migration.md` (archived work log + D135 Io strategy).
+- **W47**: `tgo_strops_cached` +24% regression investigation (single-benchmark,
+  low priority). See checklist.
+- **W45**: SIMD loop persistence — skip Q-cache eviction at loop headers
+  (requires back-edge detection in `scanBranchTargets`).
+- **W46 Phase 2 (optional)**: migrate `fd_read`/`fd_write` onto `std.Io`.
+  Not blocking. Would help return binary size to 1.50 MB target and improve
+  Windows I/O correctness further; pick up when `std.Io` stdlib APIs stabilize.
 
-- **Mac aarch64 gates**: 399/399 unit, 62263/62263 spec (0 skip), 796/796 e2e,
-  50/50 realworld, 80/80 FFI, minimal build OK, `0.16.0-baseline` bench
-  recorded (no >10% regression vs v1.9.1).
-- **Ubuntu x86_64 gates** (OrbStack): 408/411 unit (3 WAT/JIT-guarded skips),
-  62263/62263 spec, 796/796 e2e, 50/50 realworld, 80/80 FFI, minimal build OK.
-- **Branch**: `develop/zig-0.16.0` — 22 commits, ready for PR.
-- **Remaining**: PR open, CI green, close notxorand's #41, tag + CW bump.
+## Previous Task
 
-### 0.16 highlights we had to adapt to
+**W46 Phase 1c/1d/1e/1f — DONE (2026-04-25 via PR #49).**
 
-- `std.process.Init` param on `main()` (args/gpa/arena/io/env from start.zig)
-- `std.Io` threading — Vm gets an `io` field, stdlib methods take it as 1st arg
-- `std.leb` gone → inline port of 0.15's `@shlWithOverflow` algorithm
-  (`std.Io.Reader.takeLeb128` is NOT spec-equivalent; misses the "integer too
-  large" overshoot check — see `binary-leb128.77.wasm`)
-- `std.posix.*` attrition (fsync/mkdirat/dup/pread/etc.) — swap to `std.c.*`
-- `std.c.*Stat` empty on Linux — fstatat replaced by `statx` via
-  `fstatatToFileStat()`; fstat-for-size replaced by `lseek(SEEK_END)`
-- `@Vector` runtime indexing rejected → use `[N]T` arrays + `@bitCast`
-- Decisions.md D135 covers the Io threading architecture.
+Routed test-site and trace-site `std.c.*` calls through new platform helpers
+(`pfdDup2`, `pfdPipe`, `pfdSleepNs` added alongside existing `pfd*` family),
+then flipped `.link_libc = false` across every module in `build.zig` except
+the three C-API targets. CI-green on all four runners (Mac/Ubuntu/Windows/
+size-matrix). Fix commit `c11a947` routed `std.c.{pipe,dup,dup2,read,
+nanosleep}` in wasi.zig+vm.zig tests; `04ac19d` kept link_libc=true on
+C-API targets after the first push revealed `std.heap.c_allocator` needs libc.
 
 ### Hard-won nuggets (reuse later)
 
@@ -52,8 +54,15 @@ Full rundown in `@./.dev/archive/zig-0.16-migration.md` (archived work log + D13
 - **e2e_runner uses `init.io`, NOT a locally constructed Threaded io**.
   A fresh `std.Io.Threaded.init(allocator, .{}).io()` inside user main
   crashes with `0xaa…` in `Io.Timestamp.now` when iterating many files.
-
-## Previous Task
+- **C-API targets must keep `link_libc = true`.** `src/c_api.zig` uses
+  `std.heap.c_allocator`. Mac masks this via libSystem auto-link; Linux and
+  Windows fail with "C allocator is only available when linking against libc".
+- **Cross-compile sanity trick.** `zig build test -Dtarget=x86_64-linux-gnu`
+  and `-Dtarget=x86_64-windows-gnu` compile cleanly on Mac even though the
+  test binaries can't execute — the compile success alone catches link-time
+  symbol-resolution issues before pushing to CI.
+
+## Previous Task (older)
 
 **W45: SIMD Loop Persistence — DONE (2026-03-26)**