reflex-dev/reflex

reflex

Performance History

Latest Results

Drop the explicit memo body walk; document the loop-in-memo case Walking every @rx.memo body to isolate its loop items cost a compile pass per memo and had to thread the memo's own parameters into each item component, which is more surface than the case warrants. An explicit memo body is compiled as one component again and never walked, so a loop inside it renders its item inline. Per-item client state in that position is reached by making the item function an @rx.memo component, whose instances own their state; the foreach docs say so, and the Playwright scenario now exercises that pattern. Page-level loops keep rendering each item through a component of its own, so their hooks run below the item's scope. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017DE4XVgDNtq2i89CMVspJ9

claude/clientstatevar-context-refactor-jv3pig

14 hours ago

Fix hydration override mutation dependents (#7572) * fix(state): deliver hydration override mutation dependents Preserve fields-only replay output while resolving dependents invalidated by real assignments. Commit uncached delivery records against the final delta and retain shared-state cleanup bookkeeping. Fixes #7571 * fix(state): replay hydration overrides only once Mark cached storage dependents for delivery without invalidating their snapshot caches, then use one public get_delta traversal. Preserve output filters and existing synchronous/after-super ordering semantics without repeating override side effects or merging separately filtered deltas. * fix(state): share the dependent walk between invalidation and hydration replay Fold the delivery-only marking into _mark_dirty_computed_vars behind an invalidate flag, so the hydration replay reuses the one dependency walk instead of carrying a second copy of it. Move the shared-cleanup test next to the module it covers and assert that the override's correction took effect. Drop the news fragment: #7568 is unreleased, so this fix changes nothing for any released version. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_014CNQyjqBL56phnEuga6U6E --------- Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>

main

16 hours ago

docs: add news fragment for backend vars in LiteralVar

darshjoshi:fix/7459-backend-field-literal-var

16 hours ago

Refresh framework frontend pins and locked backend dependencies (#7645) * Refresh framework frontend pins and locked backend dependencies Frontend, taking the newest release at least seven days old (the window the backend lock already uses): React 19.3.1, Vite 8.3.4, PostCSS 8.5.30, postcss-import 17.0.1, Radix accordion/dialog/form/progress/ slider at the radix-ui 1.8.1 set (so they dedupe with the copy Radix Themes pulls in), react-syntax-highlighter 16.1.2, and in reflex-components-internal tailwind-merge 3.7.0 and @icons-pack/react-simple-icons 13.15.1. react-router 8.5.0 and Vite 8.3.5 are still inside the window and held for the next sweep. Backend: `uv lock --upgrade` within the existing exclude-newer window, keeping ruff and pyright at their pinned versions. Notable moves: SQLAlchemy 2.1.4, pydantic 2.14.1, Starlette 1.8.0, granian 2.8.4, FastAPI 0.143.2, python-engineio 4.14.0, playwright 1.64.0, reflex-enterprise 0.9.8. Widen reflex-release's towncrier cap to allow 26, and move the docs app to xy 0.0.8, which restores live chart rendering on Reflex 0.10. Hold the Python plotly package with plotly.js, since plotly.py 7 targets plotly.js 4, and drop the pydantic-core and pyee holds, which no longer hold anything back. SQLAlchemy 2.1 types `returning(*cols)` rows as one-element tuples; annotate reflex-workflow's lease update as a plain `Update` so its rows unpack on both 2.0 and 2.1. reflex-enterprise 0.9.8 registers a pytest plugin that imports its own var subclasses, whose generated `*CastedVar` lands in reflex_base's module; stop test_get_leaf_vars_sorts_every_cached_operation's subclass walk at classes defined outside Reflex so it only counts Reflex's own. Regenerate the docs app's reflex.lock for the new pins. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0186fG7rqdRbDhz4WvGHtvDF * Use httpx instead of requests in the URL integration test pytest-playwright 0.10 dropped its pytest-base-url dependency, which was the only thing installing requests in the integration environment, so collecting tests/integration/test_urls.py failed with ModuleNotFoundError and took every app harness job down with it. httpx is already a reflex runtime dependency. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0186fG7rqdRbDhz4WvGHtvDF * Pair each Radix primitive with its version in the news fragment Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0186fG7rqdRbDhz4WvGHtvDF --------- Co-authored-by: Claude <noreply@anthropic.com>

main

17 hours ago

fix(state): share the dependent walk between invalidation and hydration replay Fold the delivery-only marking into _mark_dirty_computed_vars behind an invalidate flag, so the hydration replay reuses the one dependency walk instead of carrying a second copy of it. Move the shared-cleanup test next to the module it covers and assert that the override's correction took effect. Drop the news fragment: #7568 is unreleased, so this fix changes nothing for any released version. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_014CNQyjqBL56phnEuga6U6E

fix/7571-hydration-override-dependents

18 hours ago

Keep the new test imports at module level The slow-consumer drain test no longer shrinks the pipe with fcntl: the default pipe already holds far less than the test writes, so the cut happens regardless, and the test now runs on every POSIX platform. psutil is imported at the top of the tree test module like elsewhere in tests. Claude-Session: https://claude.ai/code/session_012Ps5W65XWRr7LRi3c4DvQU

shutdown-owner

20 hours ago

Stop the whole tree on Windows and guard the run's cleanup window Review findings on #7650 and the Windows unit-test failure: - The JSON supervisor forwarded SIGTERM on Windows as TerminateProcess on the command alone, leaving its workers behind; it now stops the whole tree with taskkill at once, since Windows has no graceful signal. - The run's SIGINT/SIGTERM guard is installed on every platform, and the run marks the start of its cleanup so a first signal arriving during the frontend stop is ignored instead of aborting it. - Test helpers no longer race a process that exits between the liveness probe and the /proc read, the tree test checks ordering by membership, and the Windows cleanup uses SIGTERM, which exists there. Claude-Session: https://claude.ai/code/session_012Ps5W65XWRr7LRi3c4DvQU

shutdown-owner

20 hours ago

Use a Windows-safe pipe write size in the JSON supervisor select.PIPE_BUF exists only on POSIX; reading it on Windows raised in the reader thread as soon as a batch held two records. The size is now a module constant with the Linux value as fallback, the test reads that constant, and the supervisor docstring scopes the record-boundary guarantee to records that fit one write. Claude-Session: https://claude.ai/code/session_012Ps5W65XWRr7LRi3c4DvQU

shutdown-owner

20 hours ago

Latest Branches

-8%

Implement client state with useClientState hook#6936

14 hours ago

31ff217

claude/clientstatevar-context-refactor-jv3pig

+5%

fix: raise BackendVarFormatError for a backend var in a LiteralVar#7651

17 hours ago

046bcaa

darshjoshi:fix/7459-backend-field-literal-var

-8%

Fix hydration override mutation dependents#7572

18 hours ago

7c8a05c

fix/7571-hydration-override-dependents

Home Terms Privacy Docs