Regarding invalidating JWT...
Have client keep initial credentials in memory, use JWT for all transactions, have issued JWT valid for N minutes, after JWT expires, require client to get a new JWT... allow expired JWT for insensitive areas, but require reauth for sensitive ones, unless same session in the browser. By having a much shorter JWT lifecycle, you can skip the revocation server.
npm scripts have been there from the start. I've been using them as my primary entry for some while. Even with grunt/gulp, as you can simply use the version defined in package.json with a project instead of relying on a global, potentially incompatible version.
Beyond this, in general, it's just easier to work with. If you need more, a ./scripts/ directory with one or more scripts started via npm that use shelljs works very well.
Yeah, I'm definitely interested in seeing React-Native-alike options make it... IIRC there's something for the newer windows bits, but haven't looked into it, and it isn't cross-platform.
Although ESLint itself is pretty much a clean slate by default, the AirBnB config[1] is a decent starting point... and although I don't agree with everything, it's pretty understandable as a base.
[1] https://www.npmjs.com/package/eslint-config-airbnb-base
While most of the controls really don't do too much over direct HTML, I do think the notifications look decent, especially compared to the typical toast style notifications.
Didn't dig into the article, but will say in a test app I wrote a few months ago with ng2, the routing was particularly painful, getting it working with redux (syncing routing info to the state) was more painful still. And the biggest gotcha that you could only pass a route to a child one layer down, couldn't nest routes more than two layers deep.
All told, I really didn't care for ng2 as much as React+Redux, but it's definitely a step up from ng1.
All things considered not very informative of a post, but given a lot of posts targeted at beginners, it's not bad.
Personally, I think that async/await is just about the ultimate transparent use of promises.
It comes down to very different needs, and the composite fit... What would be more cool would be some kind of React Native-like binding that leverages libui under the covers... No idea if/how hard it is to develop/design custom controls against libui though.
I can't disagree more with the presumptions that are made in the description that Make is easier than JS to understand. Make requires other build tools to be installed, and depending on the environment that is complicated, if you're in a more locked down windows environment, it's downright difficult.
Not to mention that I don't like the idea of bringing in tools you don't have to. More often than not lately, I don't use grunt/gulp or the like, simple npm scripts out of package.json, and if I need more, I add them to a `//scripts` directory and call them via npm.