I'm an experienced developer and I could not wrap my head around React with free tutorials. This tutorial uses React, Flux, blah blah blah. This other tutorial uses a whole different set of packages.
It's not transferable. I need to build X. Show me a tutorial on how to build X.
People give Angular 2+ a lot of crap but at least every.single.tutorial I've ever read on it are discussing the same basic set of tools and system.
I pay for Pluralsight and while I've been focused on learning Angular from it, paid tutorials "generally" are much better quality then free ones IMO.
Well, that's a different use case.
Videos are good for exact instructionals, because of this, they are quite good for getting to the bare metal quickly and getting a quick - albeit incomplete - understanding of the technology.
Following a tutorial online on a webpage is a bit different as it's not a follow along, (you don't get told exactly what to click and "see" someone doing it) so it requires a bit more parsing and effort on your part. (Which can often be better for learning, but not always).
Generally my workflow for learning a new tech is starting with a good instructional video (like udemy) for the basic grounding, then move onto the technical documentation to develop an actual well grounded understanding.
Videos have their downsides, (you can get through an entire video series and not actually know how to self-guide) but they are much better in my opinion at lowering the barrier of entry.
Text is much better for someone wanting to learn specific things, or connecting concepts together, and/or reference documentation, searching for concepts.. etc.
The other point that I'm unsure if it is just me or all people, but I've started loving udemy videos because I can follow along with it a lot easier when I have tired eyes and worn out brain after a full day of work. I often struggle to teach myself a new technology when I'm completely worn out from my job.
However, if I get too perplexed by something and need hand-holding to understand something, I look for videos.
Also, videos help in cases with libraries whose documentations are lacking or don't have docs regarding integration with something.
Give me an in-person training course, brilliant, I'll happily watch and learn from an expert. But I just don't like video courses for some reason.
I'm actively trying to rid myself of this mindset as I recognise the benefits of using such a huge range of - really very good - resources. Maybe a 2019 New Year Resolution there!
FWIW, a lot of people tell us Redux's docs are "great", but we've also had people tell us they aren't helpful. Kinda hard to figure out how to make improvements based on that kind of conflicting feedback :)
We are actually looking at revamping the Redux docs in the near future [0]. If you've got any specific concerns or suggestions on how we can improve things, please let me know directly, or comment on that issue.
Also, besides the docs, you may be interested in my suggested resources for learning Redux [1], including my own Redux workshop slides [2].
[0] https://github.com/reduxjs/redux/issues/2590
[1] http://blog.isquaredsoftware.com/2017/12/blogged-answers-lea...
[2] https://blog.isquaredsoftware.com/2018/06/redux-fundamentals...
The problem for me is that Redux uses a set of higher order constructs, which requires a set of functional abstractions that are pretty unique to Redux and Redux-likes. There isn't remotely enough documentation to explain why the boilerplate is put where it is or named the way it is, so I have working code, but no real understanding of why the boilerplate is this way vs. any other way.
So what happens in the docs is that I follow the tutorial along, and it's great about setting up the problems to show why Redux is useful, but then it goes "ok so to solve the problem you use this magical incantation in Redux." Yes, I can adapt the magical incantation, but that doesn't really help me grok who owns which callback that makes which modifications or transformations to be consumed by which objects.
So I can totally see where conflicting feedback could appear and why you shouldn't take the "negative" feedback as a slight against the work on the docs (but fielding concerns/questions/suggestions as you are is completely a great thing).
From the outside looking in, if I were looking at Redux and I heard from a few "man, the docs are a bit hard to grasp" and others "I love the docs. They allowed me to get up and running" I wouldn't be scared off; it'd seem to be pretty par for the course.
One size doesn't fit all, and some people don't speak for all.
For me, it took me a while of working on a web app with derived data everywhere (~15 KB raw JSON that gets updated with diffs and recombobulated into different output formats and different intersections of data with live updates), and endless debugging of data transformations by stepping back and forth through the reducer history, before I really "got" it.
On the note of (b), I really like the redux-actions project (https://redux-actions.js.org) for the way its helpers that get rid of most of that. A simple example:
const setValueA = createAction('SET_VALUE_A');
const setValueB = createAction('SET_VALUE_B');
const setValueAB = createAction('SET_VALUE_AB', (a, b) => ({ a, b }));
export const reducer = combineReducers({
valueA: handleActions({
[setValueA]: (state, action) => action.payload
[setValueAB]: (state, action) => action.payload.a
}, null),
valueB: handleActions({
[setValueB]: (state, action) => action.payload
[setValueAB]: (state, action) => action.payload.b
}, null)
});There's also some helpers for common use cases like simplifying store setup, and generating entire "slices" of state at once (including generating action creators and action types automatically).
I put together an example of using this the other day:
https://www.reddit.com/r/reactjs/comments/a0b1sa/frustrated_...
Please try it out and let us know what you think!
https://github.com/reduxjs/redux-starter-kit/issues/41
Haven't had time to dig into this further, though.
Think you could leave a comment there with your suggestions and that example?
React is best to approach the same way you approach functions: You want to have small components, that do one job, that don't interact with outside state (see below for exception), and are largely ignorant of how your app works. The slight exception is container components, where interacting with outside state should be their sole job, encapsulating the application state interactions and passing the info to the contained components so that they don't have to have that knowledge.
I find a top-down approach works well: Break your page into parts (boxes of content), and then break those down, and so forth. Each box of content will be a component - most will contain others, and the smallest will just have on job.
Move all logic except for simple "show/don't show" into outside functions that you import in to your components - React is the View and you don't want it doing nearly as much as it CAN do. Keep it simple, keep it small.
Remember the key rule: You can pass values DOWN (into contained components) but not UP (into your wrapping components). So if you need to send data "up", you need to pass callbacks down and use those callbacks to pass any data "up". This can feel weird, but it helps keep those contained components ignorant of your larger application, which in turn makes them more reliable overtime, as they aren't impacted by changes outside of themselves.
If you have had limited experience with function best practices, or if you've only worked in deeply OOP systems (that have similar best practices but different concepts for passing data), this can be a bit of a struggle to adjust to, but it's actually a very common pattern, and the reason I use React to teach web dev (after teaching basic vanilla JS) - even after React is obsolete these practices will be useful to follow.
I also recommend learning React BEFORE taking on Redux (or any other application state management lib) - even a simple TODO app (don't copy what is out there, write it without hints, then compare) can help clear up the root concepts, and after that application state management is easier to keep distinct in your mind. I took it all on (React + Flux) at once when I started and it took months to grasp because I wasn't keeping them clear from one another. Redux mapXXXXToProps() functions are in fact quite hard to grasp unless you comfortable with how you use React props in practice, at which point they become straightforward.
And lastly, the switching between class-based components and function-based components can be really confusing. I personally prefer to use almost exclusively function-based components, because React is not expecting you to use an OOP approach to the components. I only use class-based components where I need component state (often only the top-level component for small apps) or if I need to use the lifecycle methods (and I rarely do - many examples use didComponentMount to fire off service calls, but I tend to put those into event handlers, though that gets a lot easier when you are using an application state management lib). but if you prefer class-based components, that's fine - just don't think React is using classes for anything more than providing inheritance of the component methods. No OOP-approach to your components.
Hope this helps!
Here are some examples:
1. Which files do I spend time on and which ones can I ignore? The first time I tried npm I felt like I had to understand all those files that kept popping up.
2. What is a good workflow? None of those documents tell you to open up dev tools and constantly console.log out values, but in a video (I assume) this will come up.
These are two examples but they mostly they show you what you should focus on. There are dozens of little things that you won't pick up from text.
Also, not going to judge a super experienced developer if they just prefer videos for whatever reason.