3 ms·

I use react/redux for a very large application and overall it has been a powerful, robust, and scalable solution that handles 99% of our use-cases. The biggest

by qudat 9y ago

I use react/redux for a very large application and overall it has been a powerful, robust, and scalable solution that handles 99% of our use-cases. The biggest problem I struggle with is knowing when components rerender. Libraries like reselect that memoize state-composed functions really help with performance, but make it less clear when things are being rerendered and why they are being rerendered. I also agree with some of the posts here that handling async actions, or side effects, can be rather difficult in the redux model. Redux-thunk is still being recommended even though Dan created it more as a bandaid solution rather than something more robust. We solved it with redux-saga, but my colleagues -- experienced developers -- struggled initially to grasp generators. I also think sagas run into the same problem that one action triggering multiple reducers has which is you could inadvertently trigger an action that hits one or more sagas and not realize you are causing side effects that you did not intent. I know there have been countless discussions on what to do about this problem, but I fear the solution will always be outside of the scope of the core library because of backwards-compat.

acemarke 9y ago
I have links to a number of useful utilities for visualizing when and why components re-rendered in the "DevTools#Component Update Monitoring" section of my Redux addons catalog [0]. Hopefully those help. I'm a bit curious why you say that "reselect makes it less clear when things are being re-rendered". The overall reasons shouldn't change - connected components will re-render when values from `mapState` or props from the parent component have changed by reference. Thunks are definitely the "minimum viable approach" for handling async logic in Redux, or at least the minimum _suggested_ approach (as opposed to doing async handling purely inside of a component). I agree that sagas (and observables) are a pretty big jump for most devs to make, but then again, async logic in general is more difficult to reason about. Personally, I still think most devs should use thunks until they see a need for something else (per my post "Idiomatic Redux: Thoughts on Thunks" [1]). Redux does deliberately push the question of "how do I handle async logic?" outside of its core, and leave it up to you. That's both because Redux focuses on structuring synchronous state updates, and because (per my post) Dan and Andrew wanted to give devs flexibility in what specific technique they use for handling async. [0] https://github.com/markerikson/redux-ecosystem-links/blob/master/devtools.md#component-update-monitoring https://github.com/markerikson/redux-ecosystem-links/blob/ma... [1] http://blog.isquaredsoftware.com/2017/01/idiomatic-redux-thoughts-on-thunks-sagas-abstraction-and-reusability/ http://blog.isquaredsoftware.com/2017/01/idiomatic-redux-tho...

by qudat 9y ago

acemarke 9y ago

I have links to a number of useful utilities for visualizing when and why components re-rendered in the "DevTools#Component Update Monitoring" section of my Redux addons catalog [0]. Hopefully those help. I'm a bit curious why you say that "reselect makes it less clear when things are being re-rendered". The overall reasons shouldn't change - connected components will re-render when values from `mapState` or props from the parent component have changed by reference. Thunks are definitely the "minimum viable approach" for handling async logic in Redux, or at least the minimum _suggested_ approach (as opposed to doing async handling purely inside of a component). I agree that sagas (and observables) are a pretty big jump for most devs to make, but then again, async logic in general is more difficult to reason about. Personally, I still think most devs should use thunks until they see a need for something else (per my post "Idiomatic Redux: Thoughts on Thunks" [1]). Redux does deliberately push the question of "how do I handle async logic?" outside of its core, and leave it up to you. That's both because Redux focuses on structuring synchronous state updates, and because (per my post) Dan and Andrew wanted to give devs flexibility in what specific technique they use for handling async. [0] https://github.com/markerikson/redux-ecosystem-links/blob/master/devtools.md#component-update-monitoring https://github.com/markerikson/redux-ecosystem-links/blob/ma... [1] http://blog.isquaredsoftware.com/2017/01/idiomatic-redux-thoughts-on-thunks-sagas-abstraction-and-reusability/ http://blog.isquaredsoftware.com/2017/01/idiomatic-redux-tho...