fix: resume InView observation when triggerOnce is disabled

[ddussi]

Hi! I put together a small fix for the lifecycle behavior reported in #781,
along with a regression test.

Summary

• resume observation in the class-based <InView> when triggerOnce changes
  from true to false
• add regression coverage for this stopped-observer lifecycle

Why

After the first intersecting entry, handleChange() invokes and clears the
stored unobserve callback when triggerOnce is enabled.

componentDidUpdate() already reinitializes observation when several
observer-related props change, but triggerOnce was not included in that
condition. As a result, changing only triggerOnce to false leaves the
mounted node unobserved until another relevant prop changes or the component is
remounted.

This change adds triggerOnce to the existing reinitialization condition. It
does not introduce a new API or a generic resubscribe mechanism.

Fixes #781

Verification

• CI=true pnpm test — 7 files, 97 tests passed
• CI=true pnpm --filter react-intersection-observer exec vitest --coverage —
  97 tests passed
• pnpm --filter react-intersection-observer typecheck
• pnpm biome ci .
• pnpm fallow audit --base upstream/main --coverage packages/react-intersection-observer/coverage/coverage-final.json
• pnpm build:all

If runtime updates to triggerOnce are not intended for the class-based API, or
if you would prefer a different lifecycle behavior, I’d be happy to adjust the
approach. Thanks for taking a look!

[bolt-new-by-stackblitz]

Run & review this pull request in StackBlitz Codeflow.

[vercel]

@ddussi is attempting to deploy a commit to the Daniel Schmidt's projects Team on Vercel.

A member of the Team first needs to authorize it.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
react-intersection-observer	Ready	Preview	Aug 11, 2026 7:12am

[thebuilder]

Thanks, this is a nicely focused fix and the reported true → false case makes sense.

One small edge case may be worth covering before merging: adding triggerOnce to the general reinitialization condition recreates the observer in both directions. A fresh observer reports the element’s current state, so toggling triggerOnce while it is already in view could repeat onChange(true) and, for false → true, immediately stop observing again.

Could you add coverage for changing it in both directions while the element is already visible, so we can confirm the intended behavior?