Repository navigation
eval: desk9 .rx reference grades 0; fix keyed refresh; tighten R-8; two guide patterns - #58
Merged
Merged
Conversation
… equations The round 7 chain4 app graded 3 defects on desk9 (fuzz seeds 8, 11, 12). The new reference keeps issue data in an app-level module and lays every write in flight over each answer through guessed sources, and grades 77/77, 38/38, 0/20. refresh of a keyed equation re-asked nothing in the compiled runtime and the oracle; S-keyed-refresh pins the fix. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
🚀 Deploying Preview to Cloudflare 🚀Preview Deployments by commit
|
…s the value Ruled 2026-11-13: pending? x takes no failure, so a region whose content reads x only under pending? never shows its fallback, and the click throws (desk9 fuzz seed 11). rxc refuses it as UNGUARDED_SNAPSHOT; the five references with the old labels.rx are fixed (labels and chain4 tasks still pass all their tests). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
Two patterns from the desk9 reference, with a generic example: writes in flight as guess lists laid over the answer by one equation every view reads, and server data plus its actions in a module so a write outlives its page. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
ryansolid
changed the title
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
ryansolid
pushed a commit
that referenced
this pull request
Nov 13, 2026
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Requested by Ryan · project thread
What
A hand-written .rx reference for desk9 that grades 0 defects on Solid
nextcd4de287: hidden 77/77, overlap 38/38, 0 of 20 fuzz seeds. Round 7's hand-written chain4 app, which has the same features, grades 3 on the same build (fuzz seeds 8, 11 and 12). The full write-up is ineval/results/desk9-reference-2026-11-13/README.md. This was prompted by Exact'sbench/reactivescript, which grades their hand-written Contract reference at 0 using this harness.Commits
eval/reference/desk9/rx/: issue data and its writes live in an app-level module. Every write in flight is a guess on its own constant-question list, laid over each answer byrowsandheading. Accepted writes end with a re-ask; declined writes revert at once.refreshof a keyed equation re-asked nothing inref/compile/rt.ts,rt-web.tsandref/oracle/lower.ts, so an action ending in one dropped its guesses. The newscenarios/S-keyed-refresh.mdpasses on all three runners.erroredregion, or adisabled, covers a handler's read of a failing value only when it reads the value itself.pending? xtakes no failure (seed 11). This is implemented in the crate, which is where R-8 has always been checked, with a unit test. The spec R-8 text,guide/reference.mdandguide/diagnostics.mdare updated. Five references with the old labels.rx are fixed; the labels and chain4 tasks still pass every visible, hidden and overlap test.guide/async.md): two patterns from the reference, using a generic example with no task code:Tests
cargo test: passes.🤖 Generated with Claude Code
https://claude.ai/code/session_0117BoWLtkg281f1NkGEDjCt