Hacker News Reader

Comment by gordonzhu | Hacker News Reader

What's most fascinating to me about this isn't the simplicity, it's that this code has lived for so long with almost no changes.

It's a great counter-example to all the recent articles about JavaScript fatigue. It's good for us to see real examples of sites that aren't caught up in the framework of the week hype.

At the end of the day we're trying to make stuff that works. ES5 vs ES6, React vs Angular vs. Ember vs. Aurelia, Angular 1 vs Angular 2, even server rendered templates vs. single page apps - it all matters a lot less than we think.

View on HN

For the majority of cases that I've seen, these arguments are simply excuses to avoid doing real work. It's way more fun to decide whether your amazingly awesome Unicorn startup is going to use React or Angular, or Express vs. Rails, or whatever than it is to actually go through the hard work of building a customer base and making money.

YES! totally agree

For a subset of us, framework authors, open source contributors, etc. It's very useful to keep up to date on the latest trends.

For example, the Angular team has borrowed techniques from React. It helps for them to keep up with things.

For people like me though, let's be honest. I don't need to do the latest thing.

Funny you said that, I was looking into ng2 and they've borrowed a few things from the React community. It's all good, I'm glad to see that "framework A is better than framework B" mentality die away and everyone just uses the right tool for the right job.

The problem (from my perspective) is that the job market follows. It's good at that--always following.

And so we use React in our project (even if it makes no sense to) because then we can put it in our resume. And thereby contribute to the problem.

Totally agree. One problem I find arises from this is that there are no boring workhorses. Time and time again I need boring crud screens, for the admin side of applications, but there is no easy default to choose. ExtJS tends to do the trick, but it's not cheap.

> Time and time again I need boring crud screens, for the admin side of applications, but there is no easy default to choose.

crud is often a feature of frameworks. crud should be an external tool like documention tools (doxygen...).

This week's JavaScript flavor special is vanilla.

maintaining software is a real thing.

for the majority of cases i've seen, these arguments are to save developers time and the company money.

my gut & feelies tell me there is truth to "it doesn't really matter what technologies you use" but my brain and experience tell me that is objectively false.

The reason "it doesn't matter what technologies you use" is because successful software gets rewritten multiple times anyway.

It matters a lot eventually, but having an "eventually" to worry about is a nice problem to have.

(There's also a big perspective difference between being an entrepreneur writing initial code for a project vs. an engineer hired to maintain & grow that code. Code quality, architecture, and technology choice matters a lot to an engineer who will be working with it directly. It matters a lot less to a business that can swap out engineers until they find ones who enjoy working on that codebase. The fact that you even have money to hire engineers usually indicates that the existing code is getting the job done. The business only gets screwed when the code quality is so bad that nobody can make sense of it, and the environment changes in a way that requires modifications.)

but, there are many instances where software would not get rewritten with proper technical foresight in the first place.

i take your point none the less- there are contexts where it doesn't matter.

Would love to see "Enterprise HN" with Node and React and Redux (with server-side rendering), and how it compares (whether horribly or awesomely).

> and how it compares (whether horribly or awesomely).

Given only the tech stack, it can go either way, depending on whether the implementor takes the easy, lazy route or actually puts some thought and hardwork into it. The problem is, as per good old Sturgeon's law, 90% of sites take the lazy route, and end up with tangled monstrosities that look like offerings to the Flying Sphaghetti Monster.

(Also, it's interesting how quickly 'cool new tech' can become 'Enterprise' shudder when it's widely abused and misused.)

> Also, it's interesting how quickly 'cool new tech' can become 'Enterprise' shudder when it's widely abused and misused

Is this the hipster corollary to coding? Sure, Java was cool when only 100 people were using it...

Well, there's Designer News...

That's already a pretty "modern" enterprise stack. I'd expect enterprise HN to run on Java EE (with an Oracle DB of course) :D

Have you seen hn.algolia.com?

https://vuejs.github.io/vue-hackernews/#!/news/1

Not exactly any of these technologies, but it is using more javascript at least. I think the example is pretty awesome actually.

Lol. For hiding a voting arrow after it's clicked. Yes, you're right. You don't need any of that stuff. But if you were to write an entire app in the style of those two functions, things would fall apart. That's not even a hypothetical, it's been proven over and over again that it doesn't work.

How is everyone agreeing with you? Is there that much JS fatigue that we're looking at the equivalent of a horse-drawn carriage, and saying wow, I wish things were as simple as they were back then?

> Is there that much JS fatigue that we're looking at the equivalent of a horse-drawn carriage, and saying wow, I wish things were as simple as they were back then?

Make Javascript Great Again! (TM)

>horse-drawn carriage

That code is more the equivalent of a digging stick. React is used to build horse-drawn carriages. I'm eagerly awaiting the JS framework that'll let me build cars.

I think the point is that there's actually little wrong with using standard HTML forms and hyperlinks most of the time.

There is an increasing trend to move the whole website to the client side to be handled entirely by the JavaScript as a single actual page.

Personally I find most websites work better with a simple round trip to the server to fetch more HTML.

So actually building an entire system in this style isn't out of the realms of possibility at all - you just only write JavaScript for the cases where there is a real legitimate benefit to doing so.

> Personally I find most websites work better with a simple round trip to the server to fetch more HTML.

> you just only write JavaScript for the cases where there is a real legitimate benefit to doing so

The snippet above only works because HN is so simple, normally the rule of thumb is if you are going to use JS to change the DOM at all then you need to render it in the first place using JS or else you are double-implementing the render logic and you open yourself up to changing one implementation and not the other. For hiding arrows sure this works but 99.999% of the time it's a very bad idea to use JS to change the dom to a state that the server would otherwise render on a page reload. You may feel that a round trip works better but that's not what users expect or will tolerate most of the time. That kind of thinking feels very close to people who think CRUD is all you need to build an app. Sure you can build an app with just that but thinks break down very quickly.

JS fatiques starts where web pages (not web apps) start to depend on thousands of kilobytes of javascript. Now the default thinking is along the lines "I am starting a new project, I will use angular", instead of some analysis what kind of project this is, how much JS is needed and what is appropriate.

I've started chiding those at work who preemptively implement features before they're required, and not scoped. Commit the simplest code that will do the job, and if you like, create a ticket for a future sprint with cool new features.

At home, I wait until lack of a feature (like hot reloading) frustrates me, then block out a couple of hours to do it specifically. Otherwise you've got a flashy stack and no product.

Nowhere did I tell people to write an entire app in the style of hide and vote like you're implying.

In my comparisons I mention a lot of choices people might make, like Angular vs. Ember vs. React that are ultimately meaningless to the success of your product. I think this is what's resonating with a lot of people.

How would things fall apart, exactly? On my home machine I browse with NoScript turned on by default, and it's rare that I find a site which is broken badly enough to notice which I still care about strongly enough that I bother clicking "allow". I don't think that most of what people do with JavaScript in the browser needs to be done at all if you filter the motivations of investors and advertisers out of the picture.

The img.src trick is used by google-analytics to log metrics cross-domain as well. Pretty neat trick.

Another thing to note is the "RESTful" nature of the links embed in the page. I guess its some form of HMAC with SHA1 + Nonce that authenticates requests for the server. It differs for every href. Elegant.

  auth=311110b7ef0f268cafa8616afcbfcec8d977111e

+1 times infinity

Reply on news.ycombinator.com