Solid 2.0.0-rc.15 Release Notes List a Fix for Concurrent-Request Hydration IDs and an Experimental Server Hook for Router Link State

Overview

The Solid project published 2.0.0-rc.15 of its packages on October 13, 2026, five days after rc.14. According to the project’s solid-js release page, the core package was tagged at 19:07 UTC that day, and the rc.14 release page shows a tag time of 02:12 UTC on October 8. The entries below come from the project’s own release notes for solid-js, @solidjs/web and @solidjs/signals; the notes for @solidjs/universal list only dependency updates. None of the entries is labeled as a security fix in the notes.

Server rendering under concurrent requests

The solid-js notes describe a fix for a context mix-up on the server. According to the solid-js release notes, an async memo’s retry now runs under its own render’s context instead of whichever concurrent render set it last (#3972). The notes say that under concurrent requests, slot content created by the retry took the other request’s hydration ids, which could leave the server spinning or fail hydration with “Hydration key miss”. The same notes say that a render effect’s retry and a projection’s re-derive, covering createProjection, createStore(fn) and createOptimisticStore(fn), now run under their own render’s context once a pending source settles.

Hydration and router link state

The @solidjs/web notes say that an async client entry can run while the browser is still parsing the server shell. In that case hydrate() gathered only the _hk nodes parsed so far, created fresh client nodes for the rest, and left the server copies parsed afterwards in the document with no listeners and no bindings (#3971). The notes say hydrate() now parks until a small marker script at the end of the shell runs, or until DOMContentLoaded if none comes. They add that a host-placed bootstrap from generateHydrationScript(), which they say includes islands hosts such as Astro, declares no ledger and never parks.

The same page describes a new experimental server API for routers, setLinkClaim(handler). According to the @solidjs/web notes, element claims never fire during server rendering, so plain anchors arrived without aria-current="page" or data-active, late on hydrated pages and never on pages that do not hydrate (#3878). Per the notes, the compilers now give every candidate <a href> in server output one hole, ssrLinkClaim(attrs), where a render’s link handler can write the anchor’s link state. On the client, hasServerLinkState() lets a link consumer tell whether the server’s anchors carry that state.

Separately, the @solidjs/web notes say element claims on the client now fire after an element’s initial attributes are applied, and that registerElementClaim takes an options bag naming which attributes re-claim an element, with a default of ["href", "action"]. The notes also remove state, noscroll, replace and preload from the core anchor JSX types, describing it as a types-only change; a routing integration is expected to declare them by augmenting AnchorHTMLAttributes.

The solid-js notes make onHydrationEnd public API again, marked @experimental (#4000). They say it is “Intended for test tooling and markers, not for branching what the app renders” and that it reports that the hydration phase ended, not that hydration succeeded.

Caching, headers and escaping

According to the @solidjs/web notes, a GET or HEAD server-function response that a shared cache may store no longer carries the request’s trace in Server-Timing, so one request’s trace is not replayed to every viewer from the cache (#3985); POST responses and documents are unchanged. The same notes say a server component that throws during its synchronous render is no longer cacheable, and that the frame response sends Cache-Control: no-store when the render fails before the head is built. They also say useHead now escapes every < in JSON script bodies as \u003C, where only </script was rewritten before.

Smaller fixes

The @solidjs/signals notes say a boundary’s fallback is now instantiated once per show. They describe an <Errored> whose content reads a failing async memo through a <Show> rendering its fallback twice for one failure and rebuilding it when a reset() retry failed again, losing focus on the Retry button (#3991). The same notes list a fix for an awaited refresh() never settling in production builds, and for affects(store) throwing a TypeError in tree-shaken production bundles (#3891).

Bundle-size figures

The release notes include size changes that the project reports for its own test scenarios. The solid-js notes put one change at -686 B minified on hydrating apps and another at -2,550 B minified on a compiled server-component page, and the @solidjs/web notes put a third at -1,165 B minified on that same kind of page. One @solidjs/web entry reports collectSlots falling from 14.2 to 5.4 ms, measured on the Hacker News story page with 652 occurrences and 17k comments; the notes describe a single page and do not say how many runs or machines were used.

What the notes do not say

The release notes reviewed here do not give a date for a stable 2.0 release. Several of the changes touch areas the notes themselves mark as experimental: the notes say the @solidjs/web/frames/server entry “is experimental” and that it no longer exports renderServerComponent, renderToFrameStream, serverComponentResponse or createFrameSink. The notes describe those functions as internal building blocks.