chore(deps): upgrade @astryxdesign packages to 0.4.0 - #2983
Conversation
Bump @astryxdesign/core, @astryxdesign/theme-neutral and @astryxdesign/cli
from 0.3.0 to 0.4.0 and rebuild the core patch against the new source.
The three 0.4.0 breaking changes miss this repo: the DropdownMenu refactor
keeps the `{type: 'divider'}` items we pass, the removed
`dropdown-menu-radio-dot` token is unused, and nothing here renders Table,
so the `useTableRowExpansion` migration does not apply.
All ten patch hunks are still needed — 0.4.0 ships none of them, including
the List `aria-label` passthrough. Only useStreamingText needed rework:
0.4.0 rewrote grapheme handling around `Intl.Segmenter`
(`snapToGraphemeBoundary`), so the `settledText` presentation state is
merged into that implementation. The reveal keeps advancing past a
surrogate pair (`advanceTextOffset`) rather than holding the code point
back a frame, which is what the repo's streaming regression tests pin;
the new snap-back then covers ZWJ and flag sequences the old hunk missed.
Verified: patch applies twice cleanly via scripts/apply-dependency-patches.mjs,
tsc --build tsconfig.lib.json, @maka/ui streaming-text and markdown-body
tests, full build, and the desktop Playwright e2e suite (22 passed).
Generated-by: Claude Code
The generated theme is a committed artifact built from the installed
@astryxdesign packages, so the 0.4.0 bump leaves it stale and CI's
`npm run astryx:theme -- --check` fails.
Two substantive changes come with the regeneration, neither of which our
own styles depend on:
- `--radius-none` is 0px instead of 0.25rem. Astryx components that ask
for no rounding were getting 4px; nothing in this repo references the
token directly.
- `.astryx-heading.display-{1,2,3}` rules now exist, so `<Heading
type="display-N">` sizes correctly under a theme that supplies
typography.scale. Our display type comes from the `--maka-text-display-*`
tokens rather than that class, so nothing here changes.
Generated-by: Claude Code
`check:release` fails the packaging job on 0.4.0: the published core tarball ships no LICENSE/COPYING/NOTICE, and the MIT copyright overrides are keyed by exact version so that a new release is re-verified instead of silently inheriting the previous version's attribution. Verified for 0.4.0 rather than copied forward: `@astryxdesign/theme-neutral` and `@astryxdesign/cli` at 0.4.0 both ship a LICENSE reading "MIT License / Copyright (c) 2026 Meta Platforms, Inc.", which is the sibling-package evidence the override comment already cites, and core's package.json still declares MIT. Regenerates THIRD_PARTY_NOTICES.txt, whose only change is the 0.3.0 → 0.4.0 version line for core and theme-neutral. Generated-by: Claude Code
|
Scoping note for review — most of this PR is mechanical and CI-backed; two parts need human judgment. 1. The merged 2. The copyright override in The remaining commits are a version bump, a regenerated theme artifact, and regenerated notices; the three 0.4.0 breaking changes miss this repo, as detailed in the PR body. CI is green 12/12. Disclosure: this change was prepared with Claude Code; the patch merge and the license verification above are AI-authored and are exactly what I'm asking you to check independently. |
English reviewConclusion: Approve. I did not find any blocking or actionable correctness/design issues. I reviewed the dependency upgrade, regenerated theme artifacts, third-party notice override, and the rebuilt Validation performed:
The exact-version license override is also supported by the published package evidence: The main residual risk is a subtle visual change caused by the upstream DOM/theme update. The successful build, Storybook/E2E CI checks, regenerated-theme verification, and class-surface comparison reduce that risk to an acceptable level.
简体中文评审结论：Approve。 我没有发现需要阻止合并或要求修改的 correctness / design 问题。 我审查了依赖升级、重新生成的主题产物、第三方许可证声明以及重建的 完成的验证包括：
精确版本的许可证 override 也有发布包证据支持： 主要剩余风险是上游 DOM/theme 更新可能引入非常细微的视觉差异。完整构建、Storybook/E2E CI、生成主题校验和 class surface 对比已将该风险降低到可接受范围。
|
The new LICENSE entry recorded theme-neutral v0.3.0 at 82d4dab, but the theme artifacts in this tree are not that build. 938487e (#2983) upgraded @astryxdesign/theme-neutral to 0.4.0 and regenerated maka.css, maka.js, and maka.d.ts with substantive output changes, and that commit is already an ancestor of this branch's merge base. The attribution was therefore wrong on arrival, not merely due for a refresh; apps/desktop/src/renderer/astryx-theme/maka.js carries `CLI: @astryxdesign/cli@0.4.0` in its generated header. Records the v0.4.0 release commit 110987b4505dc44119b94bed53d92b9840088a61 ("chore: version packages for v0.4.0", 2026-08-13), which precedes the in-repo upgrade, and syncs the current-LICENSE conclusion in the audit. Scan row 5 keeps its 0.1.0-canary.7847c5f baseline: it records what the scan compared against, not what ships. Verified: check:third-party-notices and check:cli-third-party-notices both pass; the Astryx entry sits below the THIRD-PARTY COMPONENTS header the Apache override truncates at, so no generated notice changes. Generated-by: Claude Code
styles.css told readers the "compact == transcript" assumption was held by `__tests__/markdown-rhythm-dom-contract.test.tsx`. That file was deleted in #2462 and its stylesheet-side sibling in #2425, so the comment has been citing a guard that does not exist — the worst state to leave an assumption in, since the next person reads the citation and stops looking. Restore only what cannot be noticed without a test. Both failure modes here are silent: no error, no failing check, just spacing that quietly stops being what the table declares, because a CSS selector that matches nothing never complains. - The table selects entirely on DOM Astryx generates at runtime — data-density, astryx-markdown-heading + data-level, astryx-list-item. Those names have one upstream owner and appear in Maka only inside selectors, so a rename kills every rule at once. Astryx is bumped regularly (0.4.0 in #2983, 0.4.3 in flight, plus the Dependabot minor group), so this is a recurring event, not a hypothetical. - Astryx's ListItem carries CONTROL row padding that `density` cannot reach from outside. That padding is what inverted the ladder in #2348 — list items ~10px apart against 4px paragraphs — and one rule neutralizes it. Lose the rule and the original defect returns, silently. - Plus the assumption the comment actually names: compact markdown still has exactly one caller. The table carries heading TYPOGRAPHY on a density key, which Astryx's own RFC says density must not do, so the key is honest only while `compact` and "transcript" are the same set. Deliberately NOT pinned: the ladder's declared values and their order. A first draft asserted them, and it was the wrong instinct — a reversed ladder has to be typed on purpose into four adjacent lines under a comment explaining the order, and it is visible the moment anyone looks at a transcript. It also would not have caught the defect it claimed to guard: #2348's inversion came from the ListItem padding above, which a test reading only the declared variables cannot see. Same for the adjacent-sibling gap form, the `hr` rung and the two heading size tiers — those are how the table is written, not what it promises. Why this is not what #2425/#2462 retired. Those PRs removed markup/copy pins, duplicate presentation assertions, and CSS-structure suites that asserted how a rule was written. Two of the three assertions here render the real component and check the join between Astryx's output and Maka's selectors; the third asserts an outcome the table promises, not a syntax. The stylesheet-reading half does not revive the deleted desktop css-test-helpers module or the renderer-wide CSS aggregate it read — it reads packages/ui/src/styles.css directly, in the package that owns it. Six mutations, each verified to fail the assertion that describes it: deleting the ListItem padding reset; un-zeroing it; un-zeroing it while naming the reset in a comment (this is why comments are stripped first); renaming the contract wrapper; cutting density off before it reaches Astryx; and adding a second compact caller. Generated-by: Claude Code
2 participants
Summary
Bump
@astryxdesign/core,@astryxdesign/theme-neutraland@astryxdesign/clifrom 0.3.0 to 0.4.0, and rebuild the core patch against the new source.None of the three 0.4.0 breaking changes reach this repo. The DropdownMenu refactor keeps the
{type: 'divider'}rows we pass (5 sites) rendering identically, the removeddropdown-menu-radio-dottheme target is unused here, and nothing in the product rendersTable, so theuseTableRowExpansionmigration does not apply.All ten hunks of the core patch are still required — 0.4.0 ships none of them. That includes the
Listaria-labelpassthrough:List.tsxand its build output are byte-identical to 0.3.0 and still do not destructure'aria-label', so the hunk stays.Only
useStreamingTextneeded real work. 0.4.0 rewrote grapheme handling aroundIntl.Segmenter(new exportedsnapToGraphemeBoundary; the hook signature is unchanged), so the patch'ssettledTextpresentation state is merged into that implementation rather than replacing it:advanceTextOffset) instead of holding the code point back a frame. That is whatpackages/ui/src/__tests__/streaming-text.test.tsxpins, and it is the behavior the product has shipped.Net effect for us: streaming output gets strictly better Unicode handling than either 0.3.0 + our patch or stock 0.4.0.
Verification
node scripts/apply-dependency-patches.mjs— clean, and clean again on a second run (the postinstall uses--error-on-fail, so a non-idempotent patch would break every install).node node_modules/typescript/bin/tsc --build tsconfig.lib.json— pass.@maka/ui:streaming-text.test.tsx(2 tests, includes the emoji boundary case) andmarkdown-body.test.ts(15 tests, includes thesettledTextrestore cases) — pass.npm run build— full build passes.npm run format:check— clean.apps/desktopPlaywright e2e — 22 passed, 1 skipped. Covers the surfaces most exposed to 0.4.0's DOM changes: Selector-driven settings, chat streaming (streaming-remount), quote selection, slash-command menu.astryx-*class names emitted bydist/is compared between 0.3.0 and 0.4.0 — no class disappears. The two hooks this repo styles against,.astryx-selector/.astryx-multi-selector(apps/desktop/src/renderer/styles/native-cursor.css) and.astryx-switch, both survive.Not run: the repository-wide test suite, left to CI.
CI is green (12/12). Two follow-up commits on this branch fix generated-artifact gates that the bump invalidates, each with its reasoning in the commit message:
11020338bregenerates the Astryx theme,f4d2669b0records the 0.4.0 copyright override and regenerates the notices.One
packagefailure on the way there was environmental, not ours: the job smokes our packaged 0.1.10 renderer (passed, "packaged app verified") and then re-smokes the downloaded 0.1.9 upgrade baseline, and it was the 0.1.9 binary — which carries none of this diff — that missed the 30s CDP window on a slow runner. A rerun passed unchanged.Review focus
The merged
useStreamingTexthunk inpatches/@astryxdesign+core+0.4.0.patchis the only non-mechanical part of this PR. The question worth checking is the interaction betweenadvanceTextOffset(advance forward past a pair) andsnapToGraphemeBoundary(snap back to a cluster start) — they pull in opposite directions by design, and the comment in the patch says why.Follow-ups
0.4.0 opens some cleanups that are deliberately not in this PR, since they are product changes and this branch is a dependency bump:
apps/desktop/src/renderer/styles/chat-message.css(> span:nth-child(2),> span:last-child) with the theme targets 0.4.0 puts on the icons themselves. Needs before/after visual evidence first — no test covers that CSS.MoreMenunow forwardsplacement/alignment; menu rows acceptvariant: 'destructive'and a stableid.Checklist
Does this PR entail a change in behavior?