Comment by lqet | Hacker News Reader

Comment by lqet | Hacker News Reader

Parent Full thread

Nobody ever talks about the psychological toll this takes on developers, and also on the overall software quality. If you are just re-building the same UI over and over again because "everyone is now doing it in Django / Angular / Vue / React / {insert future hyped framework}", no developer will go the extra mile to produce a stable, high-quality code base anymore. Why should they, if they instinctively know that they are just coding for the garbage bin? What you get is a team of depressed, cynical developers and low-quality software that barely works, which will then be re-implemented 12-24 months lather in some other framework, possibly by another team of developers because the old team has left in frustration or been fired for lack of motivation. It is a vicious circle that consumes the most productive age of an entire software engineer generation, and (although I am sure some of you will heavily criticize this) I am deeply convinced that the only way out of this is to just stick to basic tools and programming languages that have been around forever. Avoid any framework that is advocated like a religious cult.

Eventually, one of those current frameworks will "have been around forever". My bet is on React or one of its descendants, because viewed as a generic API it's a pretty good philosophical refinement of preexisting non-web-specific UI frameworks.

For example, there's now a framework that implements the React API for general-purpose terminal rendering, without actually involving any web stuff at all: https://github.com/vadimdemedes/ink

> My bet is on React

Exactly, it is a bet. A bet in which you can win nothing more than a refined résumé, or lose a lot of time.

In the context of web development, in the past 10 years I have seen other developers and myself write wrappers for the same JavaScript libraries again and again and again in so many frameworks and for so many CMS I lost track and count (incomplete list: typo3, Drupal, Wordpress, GWT, jQuery, AngularJS, React). This usually included weeks of work, after which it was possible to write web applications which could've been written with vanilla JavaScript on a static, handwritten HTML page in a single afternoon. From a macroeconomic point of view, it was all just an incredible waste of time. But the customers of the various companies I worked at as a student and after finishing university loved hearing the name of the frameworks. I am pretty convinced that it actually would've been enough to just write bare JS code and a HTML page by hand and sell it to them as "written in bootstrapped AngularJS". This would've saved hundreds of man hours.

> This usually included weeks of work, after which it was possible to write web applications which could've been written with vanilla JavaScript on a static, handwritten HTML page in a single afternoon.

You can have a website up and running (not just locally, but live on the Internet) within literally 5 minutes using React and Heroku, netlify, now.sh or whatever. If it takes weeks to do that, you're doing something fundamentally wrong.

It did not weeks to write the website, sorry if that was not clear from my original post. It took weeks to port existing JavaScript libraries into the ecosphere of the respective framework, to a point where it was indeed possible to write the website we wanted in 5 minutes.

> You can have a website up and running (not just locally, but live on the Internet) within literally 5 minutes using React and Heroku, netlify, now.sh or whatever.

Of course, I am not disputing this! But it is equally possible to have a website up in literally 5 minutes using vanilla JS + HTML + CSS and an nginx Server, without taking the dangerous bet that the framework you have chosen won't be obsolete in 2 years.

> without taking the dangerous bet that the framework you have chosen won't be obsolete in 2 years

That's why I specifically mentioned React and its API, as it's generic enough to have drop-in replacements or even frameworks that apply its basics to completely different kinds of rendering (https://github.com/vadimdemedes/ink). I find it unlikely that the basic idea of component trees, efficient one-way rendering of updated data, and JSX rendering will go away anytime soon, and none of that depends on the specific project named "React".

> A bet in which you can win nothing more than a refined résumé, or lose a lot of time.

I strongly disagree. Perhaps if you dont really have a problem that needs React to solve it. I understand you already had extensive expertise solving this problem in JS without React et all.

Reply on news.ycombinator.com