It was a personal choice more than anything else. I personally have more experience with Mocha and Chai and I feel like they are slightly more mature than something like Jest. As I suggested in the README, this could very easily be switched out of desired.
This repository is an opinionated repository and comes with a lot more things than the create-react-pwa. i.e. Webpack, Mocha, Flow, Nightwatch, component structure.
I agree, if you wanted to start from scratch with your own structure create-react-pwa is probably a better repository for you. However, if you like the opinions formed in this repo then it will allow you to get started straight away with static type checking, e2e testing, unit testing, tree shaking, and code splitting all built in.
You raise a good point regarding Preact, it is something that I have considered but decided to stick with React until it became problematic.
Progressive Web Apps (PWAs) are not really related to progressive enhancement.
The former is about web apps that bridge the gap between web and native. Technologically the foundation lies in using service workers for offline support, and serving the app over SSL. Not only do these web apps work offline but when using modern browsers on mobile (e.g. Firefox/Chrome on Android) it's possible to "install" them and use them like native apps.
The latter is about websites that don't rely on CSS or JS to provide critical interaction, i.e. making the website work if neither CSS nor JS is available (though generally this is simplified to "when JS is disabled").
Apart from the name the two are only related in that both are intended to help users in sub-optimal scenarios (e.g. on slow/unreliable connections).
"web apps" vs. "websites" isn't a black and white thing. SPA advocates like to say they are building "web apps" as an excuse for providing a shitty user experience. Clearly some things are more app-like (Slack) and some things are more site-like (NYT), but there's a giant gray-area in the middle. People in that gray-area call their things apps so they could avoid responsibility.
Yep. I wonder if we could memify this in an attempt to fight back: "Your f*ing blog is not an app" or something similar.
The complete abandonment of any form of progressive enhancement is depressing. We need to get some real metrics on user engagement (slow load over poor connections), SEO drawbacks (yes google can index this stuff but does it rank as well?) and accessibility (how well do screen readers really cope?).
Get some facts and start hitting people over the head with them.
You might want to look deeper into PWAs. Nothing about PWAs says you can't do server-side rendering. In fact, performance and offline support are major factors in development quality. Lighthouse (linked from the repo) checks for these things too.
Yes I suppose so - but to realistically get server-side rendering I'm currently forced to use javascript on the server as well as the client and that is not a price I'm prepared to play.
We need to get some real metrics on user engagement (slow load over poor connections), SEO drawbacks (yes google can index this stuff but does it rank as well?) and accessibility (how well do screen readers really cope?).
There's been research and data published in similar areas to all three of the suggestions I raise above. You don't need companies to do this kind of research themselves - although in some cases they do agree to publish data from studies they themselves commission.
I do see your point, but I personally like the delineation of "web app" being something that has higher modern web requirements and "website" being something that should work on something like IE9.
That is not to say that UX should be thrown out of the window for the former, of course.
I do agree that something like NYT shouldn't be a web app.
EDIT: I should mention that a lot of these things come down to requirements, of course. I don't think it's fair to say that every web app must be PWA compliant... it depends!
I don't see that such a delineation serves any useful purpose though. Does it?
The biggest argument seems to be laziness: "Progressive enhancement is hard, here's a term that I can apply liberally to give myself an excuse not to bother". That may make a good business argument on a case-by-case basis, but it's not an absolute, and I don't see why it necessarily applies to webapps but not to websites.
Also, React has made progressive enhancement a lot easier than it used to be
+1 on your point about React making progressive enhancement easier. This repo and related technologies are rather neat!
> The biggest argument seems to be laziness: "Progressive enhancement is hard, here's a term that I can apply liberally to give myself an excuse not to bother".
It's always going to come down to the requirements and an understanding of who is using your web app.
In some cases, you should absolutely optimize for PWA... Facebook, for example, is a webapp that considers users in developing countries with shotty networks. They also optimize for those with blindness.
Implementing those features in a webapp like Kibana (visualizer for elasticsearch data) would be a wasted effort given the user demographic and the purpose of the app.
TL;DR: "It depends"
EDIT: I just realized I may not have addressed your point " I don't see why it necessarily applies to webapps but not to websites.". I think the answer is simply "Webapps are inherently more complex than websites, therefore, more strict web requirements need to be met". This is an easy way to compartmentalize it, but isn't an absolute.
There is no reason these apps could not run without javascript with proper universal webapp support and declarative offline support.
My point is that they are not very progressive when they fail to run without javascript, independent of what name Google marketing team put on this bad practice.
You're mixing up cause and effect. "Ability to run without JS" is not a goal in itself. Nobody cares about users who disable JavaScript for the sake of disabling JavaScript and then complain when half the web stops working.
The reason websites should work without JS is that JS may fail to load (or worse: load incompletely or very slowly) over dodgy connections.
PWAs help by optimising repeat visits. This isn't exactly useful for something like news sites (where users typically visit individual articles at a time and may never come back for weeks/months) but can be immensely helpful for apps like Facebook where users will come back frequently.
PWAs also help by emphasising load performance. So even if the connection is slow, the app may still load quickly and reduce the time until the first meaningful paint and the time-to-interactive.
PWAs are orthogonal to making the app "work" without JS. You don't lose anything by making a regular "progressively enhanced" app a PWA. You also don't gain anything by making one that is not "progressively enhanced" a PWA.
However by using something like React to render the app you can use the same technology to render it on the server, delivering full content to the browser on the first paint. Where you take it from there is up to you and has nothing to do with PWAs or non.
As @WA suggests PWA's are JS-apps that can be used offline in a browser. You need JavaScript to instantiate the Service Worker which caches the request, without JS and in turn the Service Worker you no longer have a PWA.
PWA's are something that Google is pushing pretty hard right now, to me that suggests there is becoming less of a concern around JavaScript being turned off than previous years.
As others here have remarked, a PWA is a term with a specific meaning: it was coined with the advent of the Service Worker API to refer to any webapp using the service worker cache.
However, that still doesn't rule out producing a "progressively enhanced" progressive web app. In fact, React (which this project uses) makes progressive enhancement even easier with ReactDOMServer and the componentWillMount/componentDidMount differentiation. So the downvotes are particularly unwarranted.
I would tend to disagree with that statement. While the webpack offline plugin is creating the service worker which is putting the requests into the cache storage there is a lot more to achieving a good PWA than just a Service Worker.
You also need to reduce the Time To Interactivity (TTI) to be under 5,000ms. Webpack is doing a lot of the heavy lifting here by performing tree shaking, vendor chunking and route based chunking. All things that contribute to getting this under 5,000ms.
If you would like to know more information as to how to achieve a good PWA, I would urge you to run Lighthouse against the demo which will outline all the things that contribute to a good PWA.
No plans at the moment. I would like to make this an flourishing open source project so it may be something we discuss later on down the line. The repository is open for contributions
Personally I have kept these separate before. I would just have /src/actions/{actionType}, but, that is not to say that these cannot be included in the components folder, and actually makes sense given that a certain action is generally associated with a certain component.
* if the reducers/actions/action creators/action types/sagas are UNIQUE to the component and are not being used elsewhere then it should sit together in the same folder .
* if they are more generic/being used in more than one place then I would put them all in a separate folder, since all of those parts are actually doing more or less one thing - manage a specific sub-section of your app's state.
This sub-section of your app's state can then be shared by many more than one entry point, which is extremely useful for me.
A REAL example:
I took the redux-login-saga-flow (https://github.com/sotojuan/saga-login-flow/tree/master/app),
changed the folder structure so now i can just import the login state management into every entry point i want, and bam - it just works, since all of that state sits under 'login'.
I can now create a new webpage and add all of the login complexity in only a few minutes of work .
To be honest, you could hook this up to server-side rendering one way or another. However, I haven't looked at server-side rending too much, and I am not sure what the affect of doing this would be on the TTI. As a result, I cannot say one way or another whether this would be a good idea or not.
