refactor: avoid extra rerender

[hjiog]

Summary

useAtomValue called rerender inside useEffect, which led to an additional render, and this PR optimized that performance issue.

Check List

• yarn run prettier for formatting code and docs

[vercel]

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

Name	Status	Preview	Comments	Updated (UTC)
jotai	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Apr 29, 2024 9:35am

[codesandbox-ci]

This pull request is automatically built and testable in CodeSandbox.

To see build info of the built libraries, click here or the icon next to each commit SHA.

[dai-shi]

Thanks for the suggestion.
We can't remove rerender in useAtomValue, but you are right about the extra rerender (without commits), which isn't desirable.

I'm still working on store2, which will come with a big change, so please be patient.

That said, I think the optimization like what you suggest might be possible and meanwhile what you can help is to create a test case. Thanks for understanding.

[dai-shi]

This is required. I wonder if we could make a contrived test to cover it.

(I thought there was a rejected PR or something to refer, but can't find it now.)

[hjiog]

yeah, maybe we should add more test. but the rerender should be optimized.

[dai-shi]

Did you have a real problem in your usage? If you have, memoization might mitigate the issue for the meantime.

[dai-shi]

Please create a new test case in tests/react/optimization.test.tsx.
We can't guarantee if it will be solved.

[gurkerl83]

Thanks for the suggestion. We can't remove rerender in useAtomValue, but you are right about the extra rerender (without commits), which isn't desirable.

I'm still working on store2, which will come with a big change, so please be patient.

That said, I think the optimization like what you suggest might be possible and meanwhile what you can help is to create a test case. Thanks for understanding.

@dai-shi Could you clarify if store2 is intended to fix the underlying cause requiring a secondary render, or have I misunderstood? From your last comment about the test case, it seems this might still be an open question.

Thx!

[dai-shi]

if store2 is intended to fix the underlying cause requiring a secondary render

No, I only mentioned store2 because this PR made changes only to store2.
This behavior/problem isn't specific to store2 or store. It's working as expected.