Hojun-Cho
b26b3dbdc5
F2a batch-4 c1, mechanical, per rob's spec census (re-verified at
4adf914; check.ww/cgenstmt.ww cites were stable as predicted, the
two cgenutil [B3-SHIFT] cites moved +8: :2746->:2754, :2895->:2903).
42 conversions, every site eyes-classified plain nil+NAMED:
- cgenstmt.ww 26: one-line :149 :325 :352 :532 :551 :594 :715 :732
:938 :940 :1577 :1629 :1782 :2230 :2575 :3372 :3678 + multi-line
:986 :1912 :2287 :2291 :2702 :2759 :2823 :3037 :3715 (all plain,
no extra guards).
- cgen.ww 13: one-line :1505 :1541 :1658 :1691 :1697 :1947 :1951
:2065 :2109 :2118 + multi-line :2239 :2648 :2663 (incl. the
emitdefconstants struct/array arms — plain).
- check.ww 3: :1627 tupleelemslot, :1673 fieldslotsize, :4111 arrlit
elem chase; WHY comments kept. tichase visibility verified: same
`package wcc` (cgenstmt/cgen already call it) — no second helper.
Survivors (classified, untouched):
- cgenutil.ww :1301 tichase's own body (the accessor) + :2754
nullableptrtag single peel (batch-2 c3-B2 PROBE-CLEARED, 018ef66 —
peel-ok annotation added this commit, cites the banked record; cs
twin cmd/w6c/cgen.c:747 verified at this SHA) + :2903
flatvariantidxt pass-2 guard (batch-4 c2's site).
- check.ww :1765 comment + :1805 construction write (peellint-ok
annotation added — the one WRITE building the NAMED link) + :2038
spread peel (c3-B1) + :2273/:2275 &len/cap hand-walk (c3-B2) +
:3666 comment FALSE POSITIVE ("io.underread" matches `.under`).
- cgenexpr.ww :2886 comment only.
Lint note for the peellint finale: the `.under` token-match must be
token-bounded — check.ww:3666's "io.underread" is a comment
false-positive at a non-boundary match.
Behavior sites (:2038, :2273/:2275, :2903) and #80's site excluded
per spec — they ride c2-c4.
Byte-neutral proof: rebuilt w6c_ww on the five BASE main.combined.ww
inputs -> all five .s byte-identical to the base build's outputs;
cs==ww byte-id holds on the five REGENERATED combineds; test-unit
290 green. w6c_ww + wwdump combined.ww regens ride along.