So why is Rx and Typescript so important to me? Well specifically those two are kind of like the drug that once you try them, you literally can't go back. I've been wanting this level of type checking for 10 years on the front end. I am not willing to lose it now. As for Rx, it's the one true way to do async flows. Promises and Callbacks just end up creating large pyramidic code.
At this point it's hard to ignore that all the UI libraries are just paradigms and patterns, and how we use them is getting less important. These days I just want to write Component-style UI code, and I'll be fine.
Similarl things can be said about RxJS, but I think since it's a different level of abstraction, it will hit much harder, once it gains more traction. And it's already building up a solid community much faster than before. the only thing standing in the way of people learning RxJS is the same as Angular / Vue / React. It's scary and there's a lot of fatigue still. I hope that this year is thus going to be even better for teaching. ^^
Tldr; RxJS is going to be huge, if we can teach it more and JS fatigue calms down.
this.property$ = some_observable();
this.property$.subscribe(fn);
... will auto-unsubscribe (does it? who knows... try it?), what happens when you async pipe and its interaction with ChangeDetection...Still works, incidentally.
What _is_ interesting, though, is having your virtual DOM be the output of a reactive stream, and having a virtual DOM library use that to update the actual DOM. This is the approach Cycle.js takes: https://cycle.js.org/
We are not quite at the point of using NodeJs on the backend normally (and honestly the JVM is 100x better) but you can share the language.
You want get a django or rails like app at the moment.
I found building sites this way refreshing.
But man, am I getting jaded about all this javascript framework stuff. Every 6 months stuff breaks, every 2 years there's a huge shift.
The problem with javascript I've been facing is value. Time and effort doesn't always correlate to what I get in the other end. In fact, I can say when building, going single-page is a time sink.
And it's almost always a mistake to go SPA first. Using a django or a rails lets you get the basics and data flow nailed down early on. Get into a framework too early and have a need to change something? Have fun explaining to your manager/client how costly it is to do a "simple" modification to a JS app when you have to throw the state you built it upon out the door.
What I want is a system tightly coupled into a server-side framework like a Django or a Rails that degrades gracefully and I only have to program the interactivity one time. Something that'd plop right into the asset pipeline/django compressor so I don't have to go outside of the framework to build.
Hundreds of hours of my life have been spent chasing this dream of sharing server side code with client side JS frameworks. That's what I need.
Meteor didn't do it for me. As for rendr, I've done stuff better with backbone/express in-house. As of 2017, I get my best bang for the buck using django and pjax. No joking, I went from full DRF + Backbone Marionette -> to plain old jquery and pjax and couldn't be happier.
All these new build tools (grunt, gulp, webpack... come on), ES versions (I was ok with ES5). None of these things are helping me ship stuff ahead of / on time and correctly. They're creating an even larger gap between the server side data, logic and templates and the JS interactivity.
If anyone is listening, I'd love to have a well-supported opinionated distribution of django or rails that just renders forms, tables, etc. with angular/react/etc. and degrades gracefully.