Skip to content

Round 15 eval, catch only after await, guesses over writables, Solid main 99db9bc - #71

Merged
ryansolid merged 84 commits into
mainfrom
claude/project-thread-xp2irv
Nov 20, 2026
Merged

ryansolid merged 84 commits into
mainfrom
claude/project-thread-xp2irv

Conversation

@ryansolid

@ryansolid ryansolid commented Nov 14, 2026 •

Copy link
Copy Markdown
Member

Everything from round 15 and the board probes after it. It merges main as of #88, and every suite passes.

Language and spec

  • catch only after an await (CATCH_NOT_ON_AWAIT, R-f-4). Failures are shown in an errored region where they're read. An x = await f() catch settles the rejection where it is awaited.
  • EF-2a: an effect whose input failed doesn't run its apply. The next answer runs it.
  • AC-1g STALE_CAPTURE: after an await, an answer takes only what is the server's.
  • ARGUMENT_COUNT, props of C(), and catch reads in handlers as snapshots.
  • Guesses over writables (07 decision 21): guess works on any := source, signal or store, and on a source with a question that is also written plainly. It lowers to an optimistic layer derived from the writable. GUESS_NO_QUESTION is retired. The scenarios are S-guess-writable-1 to -3.
  • ST-4a: a key applies at any depth, as in Solid.
  • A plain write to a question source that is guessed by path no longer needs refusing, since guesses over writables now lower it.

Runtime and compiler

Solid

  • Pinned at main 99db9bce4. Covered along the way: rc.22, rc.23, and fixes #4897 through #5101.
  • Repros for every Solid bug the eval found are under eval/results/*/repro and ref/oracle/repros. Still open: #5095, #5097, #5104.

Guide

  • One shape for server data, with guesses on the answer.
  • A page's data lives in the page.
  • Two homes for a failure: the region where a read fails, and a client source the action's catch writes.
  • Region wording, with no task vocabulary.

Eval

  • Round 15: the prereg (PREREG-15, frozen), 144 runs, grades, exclusions and a provisional write-up.
  • The board task, its fuzzer (with oracles B1 to B5) and 0-defect references in all six arms. Desk9 references for React, XState, Foldkit and vanilla.
  • The board probes after round 15. On the current Solid, probe 6 grades 1, 0, 1, and both remaining defects are open Solid bugs.

Tests on this head

ref, oracle and compiled 135/135 · ssr 128 (7 skip) · types 181 · crate 372/372 · cargo test · language server 16/16 (both builds) and test/tsc.js 8/8.

🤖 Generated with Claude Code

https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
A write that lands after its page was left is not judged on the next visit
(Ryan 2026-11-14: late writes are out of scope). A4 was only applied to rows.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…XState references follow it

Ryan 2026-11-14: with two accepted writes to one field, the server's last answer is the
only correct value at rest. --r15 drops allowance A2; FUZZ_ROUND=--r15 sh harness/grade.sh.
All six desk9 references grade 0 under it (20 seeds; React, Exact, XState, Foldkit and
vanilla also 0/60 on seeds 21-80).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…erence, fuzz seed 45)

Also replay.mjs: DW=<n> widens the detail text it prints.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…ence (grades 0)

Brief frozen 2026-11-14 (Ryan). 26 visible, 45 hidden, 26 overlap tests; fuzz-board.mjs
with oracles for duplicates, headings, counts, the work-in-progress limit, the detail,
and settled state. TASK=board sh harness/grade.sh grades it. The fuzzer catches two
planted bugs in the reference (limit off by one, an accepted toggle not kept).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…el under a pending guess

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…rite visit

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…15 prereg draft

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
@cloudflare-workers-and-pages

cloudflare-workers-and-pages Bot commented Nov 14, 2026 •

Copy link
Copy Markdown

🚀 Deploying Preview to Cloudflare 🚀

Preview Deployments by commit

Status Deployment URL Commit Updated (UTC) See this deployment's details
  • Build: Failed ❌

View logs ↗
cd6eb4f 2026-11-20T19:48:47.846Z View logs ↗
  • Build: Failed ❌

View logs ↗
ca85aef 2026-11-20T08:49:55.952Z View logs ↗
  • Build: Failed ❌

View logs ↗
86d0b9d 2026-11-20T07:27:24.070Z View logs ↗
  • Build: Failed ❌

View logs ↗
d3cab00 2026-11-20T03:28:50.629Z View logs ↗
  • Build: Failed ❌

View logs ↗
dc74228 2026-11-20T03:07:26.663Z View logs ↗
  • Build: Failed ❌

View logs ↗
c3ace8c 2026-11-19T22:34:40.679Z View logs ↗
  • Build: Failed ❌

View logs ↗
f2dbd8b 2026-11-19T22:07:41.758Z View logs ↗
  • Build: Failed ❌

View logs ↗
29c9365 2026-11-19T21:54:10.528Z View logs ↗
  • Build: Failed ❌

View logs ↗
d890795 2026-11-19T17:10:23.360Z View logs ↗
  • Build: Failed ❌

View logs ↗
c2a07ac 2026-11-19T16:25:31.361Z View logs ↗

View all previews: View all previews ↗

… answered add's placeholder while work is in the air

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…, minimal trail)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…s/solid#4874, rx runtime catchRead, pending probe)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…); guide: one store over the server

- Runtime: `catchRead` with no observer (a component fn called from a handler or
  action) is a snapshot (04 R-8): the committed value, or the alternative on a
  matching failure. A fresh error boundary there had no value while a refresh
  was out and threw NotReadyError (S-catch-snapshot).
- Checker (ref/ and crate): ARGUMENT_COUNT (06 AC-1f) for a call of a declared
  action with the wrong number of arguments; `props` in `component C()` is
  UNBOUND (03 RG-4b). AST gains `noProps`. S-check-1 rejected variants, a
  check.rs unit test, guide/diagnostics.md sections.
- pending? inside a held `loading on(sel)` region reads false: Solid's, under
  05 DV-22 (plain-Solid repro saved). Guide and GUESS_WHILE_LOADING text say
  to test `pending? x` in the handler too (R-12).
- guide/async.md: "One store over the server".

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…k shown column), fuzz B3 declined count hold, repro for solidjs/solid#4887

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…on, two Solid repros

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…and by key; record seed)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…(seed 28); repro for solidjs/solid#4897

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
rx-one: a row close out holds the tab switch itself (the Help page stays until
the close answers; H-11, consistent, settles right). The ledger reference and
round 15 desk9-rx-opus-1 switch at once.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…retry as an action

Tried the two-layer board shape too (answer source + store over it, accepted
renames/assigns written into the answer, no re-ask): seed 3 lost an assign to
a list re-ask asked before it was accepted. Kept: accept = re-ask, with the
server's answer guessed on a row that wasn't on screen. Grades 77/77, 38/38,
fuzz 1/20 (seed 10, the H-11 tab-switch hold).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
Replaces 'Writing one row', 'Guessing data that depends on a selection',
'Every view of a write: lay it over the answer', 'three layers' and 'One store
over the server' with 'Guesses go on the answer' (guess on every store the
write shows in; accepted = re-ask; a row not on screen gets the answer as a
guess after the await) and 'When a guess holds'. Ledgers (writes in lists of
their own laid over the answers) are no longer taught (Ryan, 2026-11-17).
The desk9 reference in this shape is eval/reference/desk9/rx-one.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…he one latest says

latest of a list that hasn't answered throws, and the action's catch took it as a decline
(board probe 2, opus-1 seeds 2, 9, 18). Plain-Solid repros for solidjs/solid#4987, #4988.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
Probe 2: 7, 5, 3 defects, all apps on the shape. Probe 3: 9, 8, 12; the earlier-move defect is gone, holds are
longer, and an open detail-tear family grows with the extra re-asks.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
pending? is also true while a refresh or another write's hold is out (R-9);
a control under the loading boundary is enough (desk9 probe opus-1 seed 14).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…t is awaited

`x = await f() catch (e: C) => alt` lowered to an error boundary over the async
memo, so a matching rejection became an errored node first. The inner memo now
lands alt as the answer (R-f-4: "a matching rejection is an answer"); the
boundary stays only for a failed read before the await. Fixes S-fail-2 on
Solid 2.0.0 (131 -> 132 scenarios pass, S-module-1 still fails as before);
crate.ts agrees 371/372 (the other is an untracked scratch scenario) and
--lower is unchanged (190/214 before and after).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
`e catch alt` inside an equation compiled to catchRead in the equation's memo,
which made a new createErrorBoundary on every run. Each such catch (no await,
not in a closure) is now hoisted beside the equation as one catchMemo, read
where it was, and asked only while the equation's conditions reach it. This
is what hand-written Solid does, and the plain-Solid repros of
solidjs/solid#5002 and #5003 pass in this shape. Scenarios unchanged (132/1,
S-module-1 as before); crate.ts 371/372 and --lower 190/214, both as before.
spec/11 CP-4b table updated (catchMemo, settle, catchRead).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…read

R-f-4: a catch follows only an await, where a matching rejection becomes the
answer (Ryan, 2026-11-19). Any other catch is CATCH_NOT_ON_AWAIT in ref/ and
the crate. Scenarios that caught reads now use errored regions (S-fail-1/2/4,
S-module-1, S-flush-1), try/catch in handlers (S-catch-snapshot), or the error
value in an effect (S-effect-store-2). Validation is a value in the settings
example and the eval references. The guide teaches failures shown where read,
handlers that read only their own pane, and no pending? in error fallbacks
while solidjs/solid#5039 is open. The oracle lowers an await catch as the
compiled settle does. New board reference rx-regions (no catch). Repros for
#5039 and the boundary probes from the board traces.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…int answer

Drops the bullet that sent lists a write re-asks to a module with a visit
counter (written before the ruling that writes landing after their page was
left are out of scope); all three probe 4 apps followed it. Section 1 shows a
keyed store laid over a `-> null` answer so other panes never wait. Adds the
rx-regions reference grade.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…he other pane's error

Probe 5: 5 of 9 fuzz holes were a detail printing a column's failure.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…ter a failed re-ask)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
Its only evidence was one rule of the board fuzz grader (a detail must name
the card's column or show 'select a card'), not a defect a user would see.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…e's store lives with the page

An effect whose input failed no longer runs its apply with the error value:
the failure shows in the region that reads it, and the next answer fires the
apply again. Found by a desk9 outside check: an app copying a list's ids into
a source wrote the error value there and crashed on indexOf.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…ck while one has failed; guide drops board wording

Ryan 2026-11-19: a detail that must read every column fails when one fails and
should say it can't show the card right now. The reference's fallback now says
so instead of "select a card". The guide talks in regions and answers, not panes.
fuzz-board --others (exploratory) adds another user's moves and renames.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
… gone

All runners pass on it: ref, oracle and compiled 132/132, ssr 125 (8 skips),
types 178, crate 373/373. The 5 recorded divergences still diverge.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
All runners pass: ref, oracle and compiled 132/132, ssr 125, types 178, crate
373/373. The board reference's fuzz is clean on it (0/20).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…other re-asked list)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…ts list's re-ask failed)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
guess x := v is legal on any source: a := signal or store, and a source
with a question that is also written plainly. The writable is the truth;
an optimistic layer derives from it (createOptimistic(() => truth()),
createOptimisticStore(() => truth, seed)). Plain writes go to the truth,
guesses to the layer. GUESS_NO_QUESTION is retired.

Scenarios S-guess-writable-1..3. The server runtime creates the layer for
its owner slot only and renders the truth: Solid's server throws
NotReadyError for a projection over a pending store (solidjs/solid#5097,
repro in ref/oracle/repros).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
All runners pass: ref, oracle and compiled 135/135, ssr 128 (7 skip),
types 181, crate 372/372.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
…after a sibling re-ask lands)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
solid-next.sh takes #87's version (fetch the pin by hash, SOLID_BRANCH); the pin stays 99db9bce4.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
@ryansolid ryansolid changed the title eval: round 15 (prep, frozen prereg, 144 runs, provisional results) Round 15 eval, catch only after await, guesses over writables, Solid main 99db9bc Nov 20, 2026
@ryansolid
ryansolid merged commit 6a15093 into main Nov 20, 2026
18 of 19 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants