[v8] Investigate updating the React-Redux benchmarks repo

[markerikson]

Background: #1740

After v6 turned out to be too slow, we wanted v7 to be really fast. I ended up creating a React-Redux benchmarks repo at https://github.com/reduxjs/react-redux-benchmarks , which has some stress tests for creating large trees in various scenarios, and dispatching many actions quickly. It then tries to measure number of re-renders and raw update FPS as a proxy for performance.

I haven't touched that repo since early 2019. I'd appreciate it if someone could look over that repo and see what might need to be done to prepare things for working with React 18 and some React-Redux v8 alphas.

It would also be good if we can figure out more stats to record in comparison, and maybe better proxies for performance than FPS.

[ryota-murakami]

@markerikson Since Looking npm trends search result, I confirmed Recoil doesn't getting a constant user rate.

So I decide migrate Recoil to 'Redux Toolkit' as an one of acceptable de facto standard State Management library, in this project.

Create React App TypeScript Todo Example 2021

I'll rewrite this weekend and I'd like this to be used for benchmarking purposes if that is helpful for your sake.

[markerikson]

@ryota-murakami FWIW, for this issue I'm specifically interested in looking at the https://github.com/reduxjs/react-redux-benchmarks and seeing what needs to be done to get ready for updated performance testing with React-Redux v8 and React 18. Like, obviously some packages will need to be updated, but also I don't remember how the useSelector perf tests ended up being defined, or if they cover all the same scenarios that the connect tests do. I also want to see if we can improve how we're measuring "performance" here.

If you or anyone else comes up with additional benchmarks that would be useful too, but my main goal here is just the existing benchmarks repo.

[ryota-murakami]

@markerikson I see. I guess @dai-shi has some resource like that...

[dai-shi]

Well, reduxjs/react-redux-benchmarks#26 is my last work there. It should be easy to remove r-r-r, but the changes are still too big and duplicated.

If someone is gonna work on this, they can do a better way, because what we want to compare is r-r v7 and r-r v8 and comparing with connect is not primary.

[alk-mandrianarijaona]

Hello here, I am really interested in subjects around performance so I would love to give this a try. I will first take a look at https://github.com/reduxjs/react-redux-benchmarks to see how it works

[markerikson]

@alk-mandrianarijaona cool. FYI, I actually did a major update since posting this issue, to redo the build setup entirely and make it easier to add scenarios:

• Rewrite project build setup using ESBuild and a common app skeleton react-redux-benchmarks#28
• Nuke legacy CRA scenarios + counter example react-redux-benchmarks#29
• Add React-Redux v8 experimental build and initial hooks scenario react-redux-benchmarks#30

[alk-mandrianarijaona]

@markerikson thanks for that, so what are the next steps you suggest I should look into?

[markerikson]

@alk-mandrianarijaona : I just merged the one open PR I had over there . I think the next steps would be to update the React and React-Redux deps to use the latest React 18 and React-Redux v8 alphas, and then try to add more hooks-based equivalents of the existing connect-based usage scenarios.