chore(release): 0.9.4#1064
Conversation
Roll the seven publishable packages 0.9.3 -> 0.9.4 (additive release) and date the CHANGELOG [Unreleased] section as [0.9.4]. FIELD_VERSION bumped in lockstep (version.test.ts). apps/site + apps/starter version independently and are left as-is. 0.9.4 collects (all additive): - the focus / attention substrate (EXPERIMENTAL) — operator attention in, agent out - the recipe -> Pattern rename (phases 1 + 2: JS public API + Swift/Kotlin ports, with deprecation aliases) - the core folder standardization (src/core -> src/engine, math/ extraction)
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 05a6539308
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| { | ||
| "name": "@fundamental-engine/core", | ||
| "version": "0.9.3", | ||
| "version": "0.9.4", |
There was a problem hiding this comment.
This package version is the canonical value used by the native lockstep tests: Swift expects FIELD_VERSION == pkg.version (swift/Tests/FundamentalCoreTests/VersionLockstepTests.swift:34-36) and Kotlin does the same (android/fundamental-core/src/test/kotlin/com/fundamental/core/VersionLockstepTests.kt:30-33), but their constants are still 0.9.3 in swift/Sources/FundamentalCore/Engine/FieldSnapshot.swift:24 and android/fundamental-core/src/main/kotlin/com/fundamental/core/runtime/FieldSnapshot.kt:22. Since the native workflows explicitly rerun when packages/core/package.json changes, this release bump will fail Swift/Kotlin CI and would also stamp native snapshots with the old format version until those constants are bumped.
Useful? React with 👍 / 👎.
…ne lockstep) The ports carry their own FIELD_VERSION with a lockstep test against packages/core/package.json (VersionLockstepTests) — a release step beyond version.ts. Bumped both to 0.9.4. Verified: swift test --filter VersionLockstep + the Kotlin version test pass.
The one un-hardened interaction test: it fired a single pointer drag immediately, with
no guard that HomeRuntime's DEFERRED-boot [data-drag] handler had wired — a pre-boot
drag is a silent no-op (chip never moves), the chromium cold-start flake source.
- Wrap the whole gesture in expect(async () => {...}).toPass({timeout:15s}) so it retries
until the chip actually moves (same shape as the already-hardened agitate test).
- Skip the pointer path on WebKit — CI software rendering delivers pointer moves too
sparsely for the per-frame drag tracker (the limitation backlog.spec already documents);
the browser-agnostic keyboard path (also retry-guarded) covers WebKit.
Verified locally: playwright --repeat-each=3 on chromium + webkit → 12 passed, 0 flaky.
1 participant
Rolls the seven publishable packages 0.9.3 → 0.9.4 (additive release) and dates the CHANGELOG
[Unreleased]as[0.9.4]— 2026-07-10.FIELD_VERSIONbumped in lockstep (enforced byversion.test.ts).apps/site(0.2.0) +apps/starter(0.1.0) version independently and are untouched.0.9.4 collects (all additive — 0.x patch):
Patternrename phases 1 + 2 — JS public API (#1061) + Swift/Kotlin ports (#1062), with deprecation aliases (old names still work, removed at 1.0).src/core→src/engine,math/extraction.Verified:
build·version.test.tslockstep ·check:api·check:readmegreen.Publish is yours: merging this doesn't publish — pushing the
v0.9.4tag triggers the provenance publish workflow. I don't tag/publish autonomously.