[tomheaton]

Same fix as #626, for next.

saveCurrentDepth() reads window.history.state._depth right after replaceState. In an iOS WKWebView (a Flutter flutter_inappwebview embed in our case) replaceState doesn't throw but history.state stays null, so browserHistory() throws:

TypeError: null is not an object (evaluating 'window.history.state._depth')

This falls back to history.length - 1, the same value it tries to write. Nothing changes when history.state works.

Added a test that mocks history.state as null and replaceState as a no-op. It fails without the fix.

[changeset-bot]

🦋 Changeset detected

Latest commit: 4cd0965

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 1 package

Name	Type
@solidjs/router	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[ryansolid]

Confirmed on next with rc.11. With history.state reading null and the stamping replaceState silently dropped, browserHistory() throws at mount; with this change mount, push, replace and popstate all work (I ran a fuller repro with the router mounted and a useBeforeLeave registered). saveCurrentDepth was the only unguarded read — keepDepth, getSource and scroll restoration already null-check — and the history.length - 1 fallback is exactly what the stamp would have written. The only consequence in that environment is that back/forward deltas read as 0, so useBeforeLeave can't block native history there, which is the best available.

Optional nit: hoist window.history.length - 1 into a local now that it's written twice.

CI: the dist type-check needs a maintainer to approve the workflow run for a first-time contributor; nothing on your side. Will merge this; #626 (the main twin) is left for the maintainer.

— Claude via Cursor