Skip to content

Commit 697217f

Browse files
docs(plans): reader-seat rebuild §7 F-vlane — written in D-25; #3884 is the hold's own work, not a persisted posture; swap pins corrected
Co-authored-by: Cursor <[email protected]>
1 parent 7718526 commit 697217f

1 file changed

Lines changed: 17 additions & 14 deletions

File tree

‎documentation/plans/reader-seat-rebuild.md‎

Lines changed: 17 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -434,10 +434,15 @@ table) so the residue can be read against it.
434434
- **F-vlane — verdict lanes as node sets.** Today a pass that probed
435435
stages into the hold's _verdict lane_ (`CONFIG_OVERRIDE` node whose
436436
`_x._transaction` is the verdict lane), and its next pass is seated as
437-
the verdict lane's by that membership (`recompute`'s head) — even a
438-
pending re-pass whose `CONFIG_VERDICT` the head cleared (#3884), and a
437+
the verdict lane's by that membership (`recompute`'s head) — a
439438
boundary output a display-ahead re-arm swapped (`fallback`'s `b._lane`,
440-
#3540/#4337). The brief: "verdicts have NO node set; verdict passes are
439+
#3540/#4337). The #3884 pending re-pass is not among these: in
440+
`tests/superseded-transition-verdicts-3884.test.ts` the parked memo
441+
never enters a verdict lane (its probe follows a staged plain read, so
442+
`route` refuses — `verdict.ts` `JOINED|STAGED_READ`) and what the pin
443+
exercises is `core.ts`'s `CONFIG_VERDICT` head-clear; that case is the
444+
hold's own work under D-25's existing clause, not a persisted posture.
445+
The brief: "verdicts have NO node set; verdict passes are
441446
seated by POSTURE", plus D-25's probe restart. Reading A (today): the
442447
seat persists through membership — a re-pass reads held nodes as a
443448
stale reader (C) before it probes again; the swap keeps showing the
@@ -455,17 +460,15 @@ table) so the residue can be read against it.
455460
nothing, so a `Loading` fallback that went up beside the held frame
456461
stays up beside it until the landing rather than travelling with the
457462
write's frame (#3540, #4337; `tests/loading-on-latest-fallback-4337.test.ts`,
458-
`tests/loading-fallback-in-flush-3540.test.ts`), and today's verdict-lane
459-
membership is only the mechanism that happens to carry it. The general
460-
paragraph for D-25 is **not yet written**: as drafted it listed the
461-
#3884 pending re-pass among re-runs whose plain read of the held node is
462-
the screen's, which D-25's standing clause ("a pass that re-runs it
463-
while its fetch is in flight reads plainly — following the change it is
464-
fetching for — even if an earlier pass of it called `isPending` or
465-
`latest`", #3884) and `tests/superseded-transition-verdicts-3884.test.ts`
466-
contradict — the hold's own work reads its own staging, whatever its
467-
seat. Until the paragraph is re-drafted around that clause, D-25 as it
468-
stands is where the rule lives.
463+
`tests/loading-on-rearm-3540.test.ts`,
464+
`tests/loading-on-frame-following-3540.test.ts` §4–5), and today's
465+
verdict-lane membership is only the mechanism that happens to carry it.
466+
Written in D-25 as "A verdict posture outlives the pass that asked
467+
(ruled 2026-10-28)". The general persistence claim (a probing memo's
468+
later non-probing run reads a held node as the screen) rests on
469+
`core.ts` — `CONFIG_VERDICT` set at the pass tail from the probe, cleared
470+
only after a pass that did not probe, read by the next pass's frame
471+
read — and has no dedicated pin yet; S1 adds one.
469472
- **F-leaf — a leaf's children.** INV-seat-1 keeps a leaf at world S;
470473
the brief inherits `{world, lane, ahead}` from the running creator.
471474
Today a render effect that read a born node (`L4`: `joinPass`) or the

0 commit comments

Comments
 (0)