reviewer-e7 R1 finding: fold-e7 (df28784) shipped four new V-side
composition symbols (vfprintln / vfprintfln / vbsprintf / vasprintf)
but 781's rows 1-2 drove fdprintln_v / fdprintfln_v instead of the
newly-added vfprintln / vfprintfln. fdprintln_v's body open-codes
the same `match (vfprint(...)) ... match (vputbytes("\n", 1))` chain
rather than calling vfprintln, so the new V-side composition symbols
themselves had zero direct probe coverage — only transitive coverage
of the (mechanically identical) inline blocks.
Two new cstage-only rows:
- vfprintln_basic memio.fixed_vstream + fmt.vfprintln + read-back
via memio.fixed_string; asserts the 10-byte
"1 hi true\n" payload + size return.
- vfprintfln_basic same shape, fmt.vfprintfln + "x={}" + 7 → 4-byte
"x=7\n" payload.
Both rows mirror row 3 (bsprintf_v_basic) — memio.fixed_vstream wiring
+ direct vfprintln/vfprintfln dispatch + memio.fixed_string read-back —
so the new V-side symbols ride exactly the same probe shape. Cstage-
only per #209 (wwstage formattable match-arm bail); byte-id graduates
on close. Exit codes 64/65 (60-63 used by rows 1-4).
Header table updated to reflect rows 1-2's actual surface (fdprintln_v
wired sink, not vfprintln) and to cite the new rows 3-4. fmt is not
embedded in selfhost so no combined.ww regen; Makefile dep set
unchanged. 214 total tests still green (sub-rows are internal to
781_fmt_vstream_compositions_run).
Hojun-Cho
bc9d7df002