State setter functions not memoized when using NuqsTestingAdapter

[georgwittberger]

Context

What's your version of nuqs?

2.5.1

What framework are you using?

• ✅ Next.js (app router)
• ❌ Next.js (pages router)
• ❌ React SPA (no router)
• ❌ Remix
• ❌ React Router
• ❌ Other (please specify)

Which version of your framework are you using?

15.5.2

Description

When using the NuqsTestingAdapter component in unit tests (Vitest, JSDOM) the setter functions returned by useQueryState are not memoized. The function changes on every render causing effects that depend on it to rerun more often than expected.

Reproduction

https://github.com/georgwittberger/nuqs-errors-repro

https://stackblitz.com/github/georgwittberger/nuqs-errors-repro

1. Install dependencies, then run unit tests: npm run test
2. See the failing tests in src/app/setters/page.test.tsx. These tests are expected to succeed.

The component renders a text indicating whether the setter function was exactly the same as in the previous render. In these unit tests the function identity changes.

However, in the browser with NuqsAdapter from nuqs/adapters/next/app this problem does not occur. The setter function is always the same there when the state changes.

[franky47]

Thanks, the referential equality for the testing adapter should have been an issue from the start I think, as the updateUrl function isn't memoized in the testing adapter:

nuqs/packages/nuqs/src/adapters/testing.ts

Lines 33 to 39 in 29bf1fa

	updateUrl(search, options) {
	props.onUrlUpdate?.({
	searchParams: search,
	queryString: renderQueryString(search),
	options
	})
	},

Would you like to open a PR for that part?

I'll have to take a closer look at the other issue in that repro (with the transient state), it might be more tricky to fix.

[georgwittberger]

Would you like to open a PR for that part?

Not sure when I can take some time to dig into that. You mean it should work as expected if updateUrl was memoized using useCallback? And the whole context value does not have to be memoized?

[georgwittberger]

I'll have to take a closer look at the other issue in that repro

I've created a separate issue for that: #1099

[franky47]

Could you try the following preview build for the testing adapter referential stability fix please? See PR #1101.

pnpm add https://pkg.pr.new/nuqs@1101

[georgwittberger]

Fantastic, @franky47 , this works! The two unit tests for setters in the reproduction are successful with that build.

[github-actions]

🎉 This issue has been resolved in version 2.5.2 🎉

The release is available on:

• npm package (@latest dist-tag)
• GitHub release

Your semantic-release bot 📦🚀