fix(react): preserve all upstream CSS property types - #666
Merged
Merged
Conversation
The text generator only recognized single-line declarations, silently omitting five timeline trigger properties. It also coupled successful builds to whether a committed snapshot matched a moving npm latest tag. Use the exported Properties type behind the requested interface boundary so upstream declarations remain complete without a regeneration step. Constrain updates to ^0.1.36 and prove all five recovered component props with type tests, including responsive values. Six local compiler runs per version measured 0.298s before and 0.332s after on average. Preserve that measured tradeoff in the PR rather than claiming the mapped interface is free.
Contributor
Changepacks@devup-ui/components@0.1.55 → 0.1.56 - packages/components/package.jsonPatch
@devup-ui/react@1.0.41 → 1.0.42 - packages/react/package.jsonPatch
@devup-ui/reset-css@1.0.28 → 1.0.29 - packages/reset-css/package.jsonPatch
|
Codecov Report✅ All modified and coverable lines are covered by tests. 🚀 New features to boost your workflow:
|
Keep tracking upstream CSS type releases as required. Direct type consumption no longer needs a generated snapshot to match each release.
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The CSS property generator silently omits multiline declarations from csstype-extra. This restores all five missing timeline trigger props by deriving responsive values from its exported Properties type, while retaining DevupCssProperties as an interface. It also removes the generator and freshness gate, keeps csstype-extra on latest, and adds a React patch changepack.
All five new Box calls failed with TS2769 before the change and pass afterward, including a responsive array with null. The existing custom shorthand and polymorphic type tests remain intact. The empty-interface ESLint exception preserves the explicitly requested interface boundary.
The dependency intentionally remains
latestso CSS props track upstream releases. The lockfile currently resolves 0.1.36. The original locked 0.1.35 install passed the old freshness check; upgrading to 0.1.36 reproduced its stale-file exception. Consuming exported types directly removes that generated-snapshot failure while preserving the upstream update policy.Compiler measurements: Windows x64, Bun 1.4.2, TypeScript native 7.0.2, csstype-extra 0.1.36 in both cases. Six sequential wall-clock runs per implementation with PowerShell Stopwatch, before other verification builds started. Command in packages/react:
bun ../../node_modules/@typescript/native/bin/tsc --noEmit. Every run exited 0; the tsconfig includes src, not the added type-test fixture.The mapped interface was slower: mean +33ms (+11.2%), median +29ms (+9.7%). These are short local process timings, not a statistical guarantee or a consumer-scale benchmark. I consider the absolute cost acceptable for complete types and removal of the stale-file gate. If the relative regression is material for larger consumers, the next alternative to measure is an interface extending separate responsive StandardProperties and VendorProperties bases, retaining direct upstream types and the interface boundary.
Validation of the implementation before restoring the latest dependency tag (actual exit codes):
Bun: 5,180 passing tests, zero failures, 87 snapshots, and 100% reported function/line coverage. Initial attempts before building workspace dependencies failed on missing ESLint/plugin-utils build output; building those prerequisites resolved them.
The remaining raw package-lint failure is inherited: its unchanged script invokes ESLint without the TypeScript 6 preload, causing ts-api-utils to read undefined TypeFlags.Intrinsic from TypeScript 7's absent JS API. The root already documents and supplies this compatibility shim in eslint-typescript6.cjs. Supplying that same shim through NODE_OPTIONS makes the exact package lint command pass; the normal root lint command passes without extra environment settings. This PR leaves that unrelated script unchanged.
Also ran the pre-commit test command
bun run test: exit 0 (Rust tarpaulin followed by Bun tests). Local Windows Rust coverage was 98.21% (8,572/8,728 lines); this command enforces 100% only when CI is set. No Rust source changed in this PR. The normal Git pre-commit hook was retained.Previous CI passed for commit 27b7b0a (before restoring the latest dependency tag): publish, benchmark, and landing-next-e2e all passed in workflow run 34849452083, including both landing CSS modes and the CI test/coverage gate.
Restoring the latest dependency tag leaves the resolved dependency at 0.1.36. Re-ran the component type regression tests and native TypeScript no-emit check successfully (exit 0). CI for the follow-up commit will run after push.