Skip to content

Commit a7fe050

Browse files
committed
fix(signals): a boundary mounted by a pass that joined the hold appears at its commit
A born-held first pass is no longer caught by a fresh Loading boundary when the flush has joined the hold and the pass is the flush's own (not a verdict lane's mount, #3851): the mount is part of the hold, so the boundary reads its content with it and appears at the commit, its fallback never seen (D-13, U-3). This also covers a `<Show>` that opens when a flight lands: the content reads the landed value and appears with the landing instead of flashing the fallback. `recompute` passes the decision to `_fresh` (`!prevLane && flushTransaction`), so the floors pay for one argument and boundaries.ts for the test. Paid for by installing `hiddenBy` as `_hidden` directly (the `hidden` wrapper goes) and dropping two `?? 0`s before a bitwise AND. Pins flipped: loading-fallback-in-flush-3540 ":299" and the two flight shapes of "a boundary mounted by the hold". Three internal reader counts drop by the born-held memo nothing catches now. Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]> Claude-Session: https://claude.ai/code/session_011JxYhsd8CeAgCp6h5DU8ic
1 parent e371645 commit a7fe050

7 files changed

Lines changed: 134 additions & 132 deletions

File tree

Lines changed: 5 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,5 @@
1+
---
2+
"@solidjs/signals": patch
3+
---
4+
5+
A Loading boundary mounted by a pass that joined a hold appears at that hold's commit, its fallback never shown (D-13, U-3). Before, the boundary's content was caught as a fresh mount and the fallback published at once (or, for a `<Show>` that opens when a flight lands, flashed at the landing before the content). A first pass born into the flush's hold is now the hold's, not the boundary's: the content reads with the hold and the whole mount appears at its commit. A mount made by a verdict reader (`latest` in a `<Show>`) is still mainline and shows the fallback while the hold is up.

‎packages/signals/docs/OPEN-QUESTIONS.md‎

Lines changed: 11 additions & 11 deletions
Original file line numberDiff line numberDiff line change
@@ -10,16 +10,16 @@ Three sections were drafted. **Q** stays while a ruled screen is still written o
1010

1111
Each is pinned `it.fails`; the ruled screen is in `SCENARIOS.md`. Confirming here fixes the derived rule; the build order is yours.
1212

13-
| | ruling | pin | the deviation today | size note |
14-
| ------- | ---------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------- |
15-
| **U-1** | D-2 — a new async memo's answer does not show beside the old value (#3800) | `issue-3800-repro.test.ts` | built (#3977) | −22 B minified with `_bornIn` dropped (column 2) |
16-
| **U-2** | withdrawn with the direction rule (F-2). The hold waits for a mount's first load a render effect consumes outside a boundary | `direction-rule-probe.test.ts` ×3 | flipped to wait (#3977) | nothing to build. The +270 B mechanism would implement the rejected screen |
17-
| **U-3** | D-13 — a boundary mounted by a pass that joined the hold appears at the hold's commit | `loading-fallback-in-flush-3540.test.ts:299`; outside-reader form in `fresh-boundary-born-held-reader.test.ts` | it shows its fallback now, and in flight flashes it at the commit (`:299`, `:910`). The outside-reader form is not a deviation: `fresh-boundary-born-held-reader.test.ts` passes, the reader's position decides (D-13) | not measured |
18-
| **U-4** | D-14 — an outer `on`-reset boundary does not wait for a flight its nested `on`-reset boundary catches (A33, B5) | `fuzz-findings-l2.test.ts`, fuzzer finding F1 | built (2026-10-13): what an `on` re-arm collects is that boundary's alone (`arm`, boundaries.ts) | +17 B brotli / +61 B minified on the CSR scenario, within its cap (a `record` helper paid for the sweep) |
19-
| **U-5** | D-9 — a landing does not reveal a gate its stale reader re-derives onto a new flight (A15 reveal corollary) | `fuzz-findings-l2.test.ts`, fuzzer finding F2 | it reveals | not measured |
20-
| **U-6** | D-26 — a stale reader of a held re-guess shows the revealed guess, not the unrevealed one (A17, #3460) | `fuzz-findings-l2.test.ts`, fuzzer finding F5b | shows `1,2` where `1,1` is ruled | needs a per-node slot for the revealed lane value |
21-
| **U-7** | D-30 — a store row first read after an adoption derives nothing from the hold (#3706) | `adoption-unchanged-key-read-3706.test.ts:278` | built (#3986): the row is paired at that read by the family's key, as the adoption would have paired it; the pin is a plain `it` | not measured |
22-
| **U-8** | D-8 — a later mainline setter that does not write the held key publishes on the mainline tick (#3743) | `store/unchanged-presence-no-hold-3743.test.ts:491` | built (2026-10-13): a setter's written keys are its own set, the batch's earlier chained behind (`touch`, store.ts); the notify visits this setter's keys, the fold keeps the batch's union | −41 B brotli / −14 B minified on `+ createStore` |
23-
| **U-9** | D-14 — a held tree under a re-armed boundary re-derives without a redraw loop | `loading-fallback-in-flush-3540.test.ts:1020` | a redraw loop | Q-2 ruled 2026-10-09 (D-16); the loop is still the deviation |
13+
| | ruling | pin | the deviation today | size note |
14+
| ------- | ---------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------- |
15+
| **U-1** | D-2 — a new async memo's answer does not show beside the old value (#3800) | `issue-3800-repro.test.ts` | built (#3977); one shape left, a memo created inside a fresh boundary lands its first answer beside the held parent (`issue-3800-repro.test.ts:195`) | −22 B minified with `_bornIn` dropped (column 2) |
16+
| **U-2** | withdrawn with the direction rule (F-2). The hold waits for a mount's first load a render effect consumes outside a boundary | `direction-rule-probe.test.ts` ×3 | flipped to wait (#3977) | nothing to build. The +270 B mechanism would implement the rejected screen |
17+
| **U-3** | D-13 — a boundary mounted by a pass that joined the hold appears at the hold's commit | `loading-fallback-in-flush-3540.test.ts:299`, `:910` (flight, flat and nested); outside-reader form in `fresh-boundary-born-held-reader.test.ts` | built (2026-10-14): a born-held first pass is not caught when the flush has joined the hold and the pass is its own, not a verdict lane's mount (`_fresh`, boundaries.ts); the pins are plain `it` | −4 B minified on the floors, −33 B minified on the boundary apps (`hiddenBy` installed as `_hidden` directly) |
18+
| **U-4** | D-14 — an outer `on`-reset boundary does not wait for a flight its nested `on`-reset boundary catches (A33, B5) | `fuzz-findings-l2.test.ts`, fuzzer finding F1 | built (2026-10-13): what an `on` re-arm collects is that boundary's alone (`arm`, boundaries.ts) | +17 B brotli / +61 B minified on the CSR scenario, within its cap (a `record` helper paid for the sweep) |
19+
| **U-5** | D-9 — a landing does not reveal a gate its stale reader re-derives onto a new flight (A15 reveal corollary) | `fuzz-findings-l2.test.ts`, fuzzer finding F2 | it reveals | not measured |
20+
| **U-6** | D-26 — a stale reader of a held re-guess shows the revealed guess, not the unrevealed one (A17, #3460) | `fuzz-findings-l2.test.ts`, fuzzer finding F5b | shows `1,2` where `1,1` is ruled | needs a per-node slot for the revealed lane value |
21+
| **U-7** | D-30 — a store row first read after an adoption derives nothing from the hold (#3706) | `adoption-unchanged-key-read-3706.test.ts:278` | built (#3986): the row is paired at that read by the family's key, as the adoption would have paired it; the pin is a plain `it` | not measured |
22+
| **U-8** | D-8 — a later mainline setter that does not write the held key publishes on the mainline tick (#3743) | `store/unchanged-presence-no-hold-3743.test.ts:491` | built (2026-10-13): a setter's written keys are its own set, the batch's earlier chained behind (`touch`, store.ts); the notify visits this setter's keys, the fold keeps the batch's union | −41 B brotli / −14 B minified on `+ createStore` |
23+
| **U-9** | D-14 — a held tree under a re-armed boundary re-derives without a redraw loop | `loading-fallback-in-flush-3540.test.ts:1020` | a redraw loop | Q-2 ruled 2026-10-09 (D-16); the loop is still the deviation |
2424

2525
**Answer format that is enough:** `U-1, U-3–U-9 stand, U-6 after 2.0` — or the line you prefer. U-2 is withdrawn with the direction rule's time half (once drafted as F-8; F-2). Q-1 through Q-5 are ruled and carried by the rules. C-1 is D-12. C-2 is D-7. C-3 is D-32.

0 commit comments

Comments
 (0)