reflex-dev/reflex - CodSpeed

reflex

Blog Docs Changelog

Performance History

Latest Results

Add changelog news fragments for #6850 One fragment per affected package: the user-facing as_child/Slot fix for reflex, the mergeSlotProps/mergeRefs wrapper mechanics and the mergician base dependency for reflex-base, and the unversioned mergician import for reflex-components-plotly. Verified with towncrier draft builds against each package directory. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01TgUwZh63X1rWaYrfExH2Hg

claude/automemoized-forwardref-protocol-w2fe3p

13 minutes ago

Drop the mergician version specifier from the plotly import With mergician in the base package.json dependencies, plotly's pinned "mergician@v2.0.2" import key no longer collapses during package collection: _get_frontend_packages filters page imports against base dependency names by exact string, so the versioned key would install separately and could drift from the base pin over time. Importing "mergician" unversioned resolves to a package name that matches the base dependency and is filtered out, leaving installer.py as the single owner of the pin. The compiled JS import is unchanged either way. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01TgUwZh63X1rWaYrfExH2Hg

claude/automemoized-forwardref-protocol-w2fe3p

16 minutes ago

Stabilize composed ref identity and deep-merge object props via mergician mergeSlotProps runs inline in the wrapper's JSX, so composing refs with a fresh mergeRefs callback per render made React detach and reattach the refs (running callback cleanup or transiently nulling object refs) on every rerender of a mounted wrapper even though the node was unchanged. Composed refs are now cached in a WeakMap chain keyed by the (own, injected) ref identities: stable inputs reuse the same composed callback across renders, and non-weakable inputs fall back to uncached composition. Object-valued props (style, css, and other nested config) now deep-merge per-key via mergician — as elsewhere in the codebase — with own keys winning, instead of an own object dropping every injected key wholesale. Plain objects only: arrays, React elements ($$typeof), and class instances keep own-wins-wholesale semantics. mergician joins the base package.json dependencies, pinned to the same v2.0.2 the plotly wrapper requests. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01TgUwZh63X1rWaYrfExH2Hg

claude/automemoized-forwardref-protocol-w2fe3p

43 minutes ago

Compose refs inside mergeSlotProps and drop the compile-side ref pass With ref-as-prop in React 19, the wrapper's ...rest already captures an injected ref, so the transparent signature shrinks to ({children, ...rest}) and ref composition moves into mergeSlotProps: a ref present on both sides merges via mergeRefs at runtime, the root's own id-derived ref otherwise stays a plain object ref with no wrapper callback allocated per render. This deletes the compile-side _forward_incoming_ref_to_root pass entirely — the root's own ref already rides its rendered props. Also drop the style special case from mergeSlotProps: compiled output carries styles on emotion's css prop (style is a Python-side concept), so an injected style prop cannot collide with own props and passes through the base spread untouched. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01TgUwZh63X1rWaYrfExH2Hg

claude/automemoized-forwardref-protocol-w2fe3p

1 hour ago

fix: reload once on frontend/backend state mismatch A stale frontend build is the usual cause of a delta the frontend cannot dispatch, and a reload picks up the matching one. The reload is recorded in sessionStorage so a mismatch that survives it (e.g. api_url pointing at a different app) leaves the page up with the error reported rather than looping. Also from review: - Move _sanitize_client_log_value off EventNamespace into format.sanitize_client_log_value, slicing to max_length before the per-character scan so an oversized value costs no more than a bounded one. - Sanitize only after the SID and rate-limit checks, so reports that get dropped do not pay for it. - Escape rich markup in FrontendEventExceptionState, where a JS message containing square brackets could style backend logs or raise MarkupError. - Validate the delta in a single pass that allocates only on a miss, and stop awaiting queueEvents inside the event handler so a delta applies in full before a later update can interleave. - Fold tests/units/test_client_error.py into test_app.py and test_format.py, and cover the report and the reload in Playwright.

FarhanAliRaza:fix/unprocessable-delta-client-error

2 hours ago

Merge Slot-injected props onto auto-memo wrapper roots Refs now forward through generated memo wrappers, but props a parent injects at runtime were still dropped: a Radix asChild/Slot parent clones its child element and injects props onto it, and the wrapper's destructured signature discarded everything it didn't declare. The user-visible symptom (reflex-dev/reflex#6849) was a stateful input under rx.form.control(as_child=True) rendering with no name/id, so new FormData(form) omitted the field entirely. Transparent wrappers now compile to ({children, ref, ...rest}) and render their root as jsx(Root, {...mergeSlotProps(rest, ({...own}))}). The new mergeSlotProps runtime helper applies Radix Slot merge semantics with the compiled-in props in the child role: own props win for plain props, on* handlers compose (own first), style objects merge, and className strings concatenate; an empty injection returns the own props object unchanged. The forward_root_ref definition flag is renamed to forward_root_props to cover both halves of the transparency. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01TgUwZh63X1rWaYrfExH2Hg

claude/automemoized-forwardref-protocol-w2fe3p

2 hours ago

Forward refs through auto-memoized component wrappers Generated memo wrappers destructure only their declared props, so a ref set on the wrapper at runtime (e.g. injected by a Radix Slot/asChild parent cloning its child element) was silently dropped and never reached a DOM node. Wrappers built by create_passthrough_component_memo whose memo body renders a tagged root now implement the React ref-forwarding protocol: the compiled function destructures the incoming ref prop (React 19 ref-as-prop) and attaches it to the root element, composed with the root's own id-derived ref via a new mergeRefs runtime helper that supports ref objects and callback refs, including React 19 cleanup returns. Untagged roots (Bare, Cond, Match, Foreach) and Fragment roots are unchanged, as are user-defined @rx.memo components, which keep their documented RestProp contract. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01TgUwZh63X1rWaYrfExH2Hg

claude/automemoized-forwardref-protocol-w2fe3p

3 hours ago

Bump base-ui version to 1.7.0 for reflex components internal (#6842)

main

6 hours ago

Latest Branches

0%

ENG-10883: Make auto-memoized component wrappers transparent to injected props and refs#6850

16 minutes ago

936c324

claude/automemoized-forwardref-protocol-w2fe3p

0%

feat: surface unprocessable state deltas as fatal client errors#6827

2 hours ago

8564cb6

FarhanAliRaza:fix/unprocessable-delta-client-error

-3%

add orjson#6116

1 day ago

bcd6574

benedikt-bartscher:try-orjson

© 2026 CodSpeed Technology

Home Terms Privacy Docs