Hacker News Reader

Comment by gfodor | Hacker News Reader

I guess I'll never really understand articles like this, or I just have poor reading comprehension. I don't see what the point of this type of thing is, it feels like a rant without any particular recommendations. It can be summed up as "don't use technology without understanding it, and don't blindly follow the crowd." But then again, how are you supposed to learn when a certain piece of technology should be applied, other than apply it and see how it goes? I certainly am not going to just take someone's advice and opinions at face value, unless they can provide hard evidence of why a certain technique results in failure. It requires a certain amount of cognitive dissonance to point to Etsy's 2000 files as a failure while at the same time they are a successful company. Perhaps these bits of incidental complexity we like to obsess over aren't, in the grand scheme of things, all that important?

To me, the notion of having strong opinions on particular techniques for web application (or web site) development strikes me as an example of a subfield having an over-inflated sense of self. Web app development and javascript frameworks are a tiny, tiny niche of software engineering which is an even smaller niche of the entire area of the medium of computing. And it's not a particularly interesting one, imho. I'm not sure why so much ink is spilled on it other than it's easy to form opinions around.

View on HN

> I'm not sure why so much ink is spilled on it other than it's easy to form opinions around.

Because people read it. And people read it because social cues are important.

When you're going to commit hundreds (or thousands) of man hours to building a system you may have to maintain for years based on very, very partial knowledge, you want to know what other people think about the tools you plan to use.

> It can be summed up as "don't use technology without understanding it, and don't blindly follow the crowd." But then again, how are you supposed to learn when a certain piece of technology should be applied, other than apply it and see how it goes?

That's called research. You do it on your own time, while using more mature & proven solutions for daily work. Play with it, review all the source code. Like it or not, it's going to be your code, once you start using it in earnest. Don't treat it like a friendly black box.

Is that too much to do? Then you're going to have a lot of surprises. Maybe that's fun for you, but I have to say it gets really old over the years.

> It requires a certain amount of cognitive dissonance to point to Etsy's 2000 files as a failure while at the same time they are a successful company.

Er, no, it doesn't require cognitive dissonance at all. In fact, the folks at Etsy blogged about the issues themselves.

It's a common thing for a fast moving, highly successful project to accumulate gunk under the excuse of "it works for now, we'll clean it up later". But, if the clean up never comes, that gunk eventually becomes a noticeable drag on daily development.

> And it's not a particularly interesting one, imho. I'm not sure why so much ink is spilled on it other than it's easy to form opinions around.

If it's not particularly interesting, then why are you reading about it and responding to it? :)

Maybe I can fill in the gap?

If you are a developer, you shouldn't rely on Javascript. Yes, feel free to use Javascript to make things sexier, but the use cases where a web page should require Javascript to render properly has substantial overlap with the "we didn't think things through" slice of the Venn diagram.

If your page is unreadable with NoScript and RequestPolicy turned up to maximum paranoia, your webpage is broken.

If you're building a client-side application rather than just a web page, then it's okay if turning JS off breaks your app.

I think I gathered this, but my point is why does this matter? If I run a site that is unreadable on NoScript and RequestPolicy turned up to maximum paranoia, but am generating millions of dollars of revenue, is my webpage "broken?" If only 0.01% of my users actually are affected by such a thing, and I don't care about that, is it still "broken?" If a webpage has a missing closing </div> tag, but is never hit by a human being, is it still broken?

I guess what I'm getting at is as I've gotten older I've grown less and less empathetic to arguments that seem to appeal to an engineer's sense of purity, correctness, and craftsmanship and not much else. Tie it to user pleasure, effectiveness, business metrics, anything, but citing the idea that somehow since the web was not designed with Javascript everywhere in mind meaning we should use it as the exception not the rule isn't much more than appeal to tradition argument to me. There are definitely areas of software engineering where you can find many examples where a lack of respect for the craft, or a lack of knowledge of history/standards results in failure, but I'm just not buying the idea that if you build a SPA with a fancy new javascript framework instead of making sure NoScript works you're somehow going to be paying for that in spades down the road because it's breaking some set of rules handed down by the high priests. There are about a million other technical decisions that have more of an impact in a serious project. (For example, instead of arguing about SPA vs non-SPA, or what framework, or whatever, ask yourself: "Is anyone on the team an expert in any particular flavor of this?" If the answer is "yes", then going with any one of those is almost certainly the best strategic decision regardless of the supposed merits or flaws of that particular approach.)

> If only 0.01% of my users actually are affected by such a thing

All users don't have the same value. And it's also about trends.More and more users use addons that block javascript by default,just like more and more users use adblock. You might not care today,but when it's time to care, well your competitor that cared will get the users and the revenue lost because "you didn't care".

> And it's also about trends.More and more users use addons that block javascript by default,just like more and more users use adblock.

Do you have any data to back that up? In my experience it seems like the number of people who turn JS off by default has declined since SPAs became so common. It looks like the search trend for Noscript has been gradually declining since 2009, while Adblock has risen: http://www.google.com/trends/explore#q=noscript%2C%20adblock...

This should have a huge *citation needed. There are a million decisions you make which affect user adoption and retention of a software product, deciding to support NoScript users is a small grain of sand in that desert. Your claim that this facet of all those decisions will result in a competitor stealing market share completely ignores the concept of engineering tradeoffs. The competitor that "cared" about noscript visitors may spend many man hours (in direct costs or opportunity cost) supporting those visitors, time may be best spent building better competitive differentiators for a larger number of visitors. No design decision lives in a vacuum, and in my own experience designing to support non-JS visitors is almost universally faaaar down the list of priorities if there are good reasons to not do so.

> the concept of engineering tradeoffs.

supporting no script is as simple as not abusing javascript in the client.I fail to see how you'd spend more man hours when not investing in these full client-side javascript frameworks when your product can work without using javascript to begin with.

See Gmail for a strong counterexample. It's UX is fundamentally about speed and responsiveness, and would have been difficult to build as one of progressive enhancement, hence they needed to build a completely separate user interface to support non-JS users.

>0.01%

13%, actually, globally. 2% in the US (population of NYC):

http://www.searchenginepeople.com/blog/stats-no-javascript.h...

That doesn't even count people who use browsers which you haven't tested with obscure bugs, or people who have laggy and slow connections that also breaks or damages the functionality of your javascript heavy web app.

The point wasn't the number, the point was if only a small % of my users are affected by a given design decision, why should I care? Beyond that, website populations have bias and as such citing national/global stats is at best a first order approximation of the impact of not supporting non-JS visitors.

>why should I care?

I explained why. For people who use browsers who you haven't tested which break the JS in obscure ways, not to mention people whose experience breaks in obscure ways because of dropped packets and high lag.

>Beyond that, website populations have bias and as such citing national/global stats is at best a first order approximation

Ironic complaint from somebody whose first guess was two to three orders of magnitude off.

2% of my users can have javascript turned off and yet only 1% of those people may be negatively affected by the decision to not ensure all features of my site work without it.

but go on, keep missing the wider point by being pedantic.

Plus, it's just railing against the tide. I don't disable JS. None of my friends disable JS. (In fact, most of my friends don't even know what JS is.) The setting to disable JS is hidden away in some obscure pref pane in most browsers. It's not going anywhere, it's not optional, and pretending that a site is "broken" if it doesn't work for a handful of geeks is being willfully obtuse.

It's worth noting that "disabling JS" is not the only way to disable JS. Sometimes it can happen inadvertently:

* Bug accidentally makes it into production, preventing some crucial feature from working? JS just got disabled.

* User is browsing your site with an old device that doesn't support a feature you make use of? That JS has been disabled.

* CDN fails and your fallback mechanism isn't working like you thought it was? JS got disabled.

Not saying these are common occurrences, but I've seen all of them happen.

Actually, didn't Firefox already remove the checkbox? It's probably now only visible under about:config...

The big reason to avoid shipping a white page with JS turned off is performance. It implies that at some point, the JS will load and you can stop using ng-cloak or whatever else is hiding in the blank space.

It's better to pre-load actual content, or if you don't have that, fake content, then to show nothing at all.

Similarly, it's better to pay attention to back buttons so your users can navigate a website using built-in controls. This is especially important on phones, where it's easy to click a wrong link, and where back is literally a button.

Can you do this with JS? Yes. Is it obvious? Not always. The same was true with our use of CSS over images, or web fonts over flash. Eventually the industry will move in the right direction, it just takes time and the occasional rant or two. ;-)

How about screen-readers and other assistive software for visually impaired users? Apparently they've gotten better in recent years, but if you make a website that doesn't work without JavaScript, you're still probably screwing a lot of those people over.

Maybe they're a small percentage of your users, but it still strikes me as a dick move in the same category as constructing a building that isn't handicap-accessible (yes, I'm aware there are legal compliance rules in the building trade).

It's mostly a myth that relying on JavaScript makes a website inaccessible (partially propagated by the woefully out of date Section 508 requirements in the US - the update is currently in progress and will be very close to WCAG 2.0 [1]).

The reality is that very few screenreader users disable JavaScript [2] surprise! they're much like any other user). A properly coded SPA, using aria-* properties is as accessible to most AT users as a traditional server-side application.

[1] http://www.w3.org/TR/WCAG20/

[2] http://webaim.org/projects/screenreadersurvey5/

> If I run a site that is unreadable on NoScript and RequestPolicy turned up to maximum paranoia, but am generating millions of dollars of revenue, is my webpage "broken?"

I'd say yes. Usable != profitable. Hey, HN is broken, because it will "time-out" pages and you will sometimes get an error when clicking "next" - but that doesn't make it useless. (and for the arc apologists that usually respond to this - I clicked a valid link on this website, I don't care what technology is HN based on)

Also, your website would be broken if it was unusable: under anything other than IE, on machines other than PCs, on devices not made in Asia, by people older than 60, etc. None of those prevent it from generating millions of dollars.

Then again "broken" doesn't mean you should care. If your goal was just to make millions of dollars, that's fine. Just don't be surprised people call it broken.

See, by that standard the word "broken" loses any significant meaning and becomes noise. Why care about anyone calling anything broken then?

I don't think it's meaningless. It means you're making $X, but ignoring N% of potential audience. If you fix your stuff you could potentially reach them and make $X+Y.

Outside of websites: Private healthcare is broken in the US, but works for N% of people who can afford it. Many people live with the status quo. Is it meaningless to call that system broken? If it's a high enough percent, why care about anyone calling it broken? [my answer - because we're aware it could be better, but it requires more work]

>If you're building a client-side application rather than just a web page, then it's okay if turning JS off breaks your app.

I wouldn't even say that this was necessarily true. Client side applications that can be made perfectly functional without JS should be made functional with JS.

I don't want my online banking website or my tax return web apps to use javascript at all. They do, though.

The fact that people make blogs, whose only real function is to serve static text, render with javascript is nuts.

Why should your bank spend more money and effort coding and testing both a JS enabled version plus a non JS version?

The idea of no JS gets worse when you start applying the idea of just making an API, then having the client just use those APIs, whether they're browsers, backend systems or mobile clients.

Requiring two separate client UIs to accommodate no-JS is unlikely to be a worthwhile cost for many sites/applications.

>Why should your bank spend more money and effort coding and testing both a JS enabled version plus a non JS version?

My bank should not create a site that requires javascript at all.

It increases the attack surface, it increases the likelihood that there will be bugs lurking and it increases the likelihood that the site will fail under circumstances where the user has low bandwidth and lots of dropped packets.

Furthermore, it is not necessary for a working site that does everything that I want. At all. So what is the point??

>The idea of no JS gets worse when you start applying the idea of just making an API, then having the client just use those APIs, whether they're browsers, backend systems or mobile clients.

Which I would not advise doing under any circumstances. Sharing common REST APIs between browser, backend systems and mobile clients sounds like a great idea until you consider that they all require different modes of authentication, authorization, access policies and security features and the like. Are you going to force your backend systems to authenticate using cookies and use the same CSRF protection?

Having separate APIs need not mean that you duplicate a lot of code, so I don't see the need to try to accommodate multiple different use cases using the same APIs.

>Requiring two separate client UIs to accommodate no-JS is unlikely to be a worthwhile cost for many sites/applications.

Requiring JS for many sites/applications is unlikely to be a worthwhile cost, especially considering the additional testing required to make it work equivalently well to a non-JS enabled site/application.

In general, the thing that draws people to building a JS-driven site is because it allows faster responsiveness for users, and a more versatile interface that behaves more like a application and less like a document. I think by now this much should be obvious. If the standard HTTP request/response approach resulted in the responsiveness in feedback we see with AJAX, then we wouldn't have been having this discussion for the last 10 years.

So in short, if a JS-oriented banking website results in a user being able to do their banking more quickly, this is a real benefit and is "the point." You can argue it is not worth the tradeoff, but it's a bit odd to first admit you don't see the point and then go on to say it's not worth it. If you don't see the point, how can you make claims about if it's a valuable thing to be doing? If a bank has 50k daily active users and it shaves 1-2 minutes off of their individual session times, you're looking at saving approximately 50 days of time for people every day. And most banks probably have way more than 50k daily actives. Combine this with other potential benefits like users being able to get enough done in a single banking session to avoid a second visit later, or the user interface being less confusing since users can fall back on desktop metaphors, the amount of saved effort could become truly staggering, worth the additional risk and engineering time of developing a SPA-based solution.

>In general, the thing that draws people to building a JS-driven site is because it allows faster responsiveness for users

The reality, however, when you create sites like banking websites is that it is more often slower.

Google maps is the exception, not the norm.

>a more versatile interface that behaves more like a application and less like a document. I think by now this much should be obvious. If the standard HTTP request/response approach resulted in the responsiveness in feedback we see with AJAX, then we wouldn't have been having this discussion for the last 10 years.

We've had the NoSQL discussion for an equivalent duration. Ironically, the source of this discussion was the same in both cases - that because it worked well for Google's specific requirements, it should be used everywhere. That proved to be an unfounded assumption.

>So in short, if a JS-oriented banking website results in a user being able to do their banking more quickly, this is a real benefit

Unfortunately, it doesn't usually result in a faster website. The banking website I use that makes heavy use of javascript is actually slower than those that do not.

>You can argue it is not worth the tradeoff, but it's a bit odd to first admit you don't see the point and then go on to say it's not worth it.

The point was fairly clear: unless you can make a significant speed gain (and more than you think can not), then there is no trade off. It is universally a bad idea.

>If a bank has 50k daily active users and it shaves 1-2 minutes off of their individual session times, you're looking at saving approximately 50 days of time for people every day.

That's the second time on this thread that you've pulled a number right out of your ass.

>Combine this with other potential benefits like users being able to get enough done in a single banking session to avoid a second visit later, or the user interface being less confusing since users can fall back on desktop metaphors

Then there are those AJAX websites that break the back button. Users LOVE that.

Why do you keep focusing on the numbers in my hypothetical examples? If you've run a large enough website, with enough users small optimizations end up being magnified into real effects. That's the point, and you're missing it again by being pedantic.

Beyond that, specific claims like "when you create sites like banking websites is that it is more often slower" are things that require evidence, and are much more in the category of 'pulling something out of your ass' than me very clearly constructing reasonable hypotheticals, not making claims about what would happen on a specific site or situation.

fwiw: I have run lots of A/B tests and in many cases converting things to AJAX-y, less HTTP-y solutions increases metrics meaningfully (like being able to add things to a list in a single click, instead of a full HTTP POST), and in many cases it does not (like in cases of infinite-scrolling through results.) Go figure, context matters and making flat-out claims is dumb. It's perfectly reasonable to believe there is some significant chance that a highly interactive site like a bank could see real meaningful changes in user behavior if it avoided doing full HTTP postback for all its pages.

So if my bank didn't use JS, the already drawn out processes would take even longer. Transfers would require more clicks, more full page reloads. No benefit to the user. In fact, the opposite. The user just feels their banking system is even older and more outdated. So your use of "not necessary" is rather debatable and probably not the same as the PM's.

As far as security on the API, I'm not convinced. Don't use cookies for authentication -- is that a big problem?

> If you are a developer, you shouldn't rely on Javascript.

I'm sorry but I think he's saying something a little to the left of that. His overarching point about javascript seems to be that it adds complexity to any application - suddenly, page loading indicators and all application states must be handled intelligently by us, developers. His point here is that we already have an issue with developers working on code they don't have the experience, ability or time to understand. SPAs and anything else lead to a serious requirement for developer discipline.

I think some important points were, in my own words:

- Choosing a technology stack doesn't matter

- Choosing a framework doesn't matter

- What the application does and how it behaves is all that matters. Choose any tech stack that will do the job.

- No matter what tech stack you choose, your app will run into issues. It will need maintaining.

Like he says:

> What a good app needs is a team of dedicated people behind it. Good apps are good because talented people cared about them and pooled their skills.

- No amount of abstraction will help. If the abstraction is not understood, it will also hurt.

But please feel free to agree or disagree with me. I'd love to know what you all thought.

Choosing a tech stack and framework DOES matter greatly. Some technologies make development a lot easier and maintainable. Otherwise everyone will still be using PHP to write their websites. Abstractions does help greatly and frameworks are a good way to get your entire team to think about things using the same abstraction.

The platitude about needing a "talented", "passionate" team is vacuous and meaningless. You might as well write the key aspect of a good app is very little bugs. Absolutely meaningless statement. It's very obvious a bad developer with a good framework will lead to bad output; no need for a 5000 word article on it.

(Almost) everyone is still using PHP to write their websites. That has nothing to do with front-end frameworks. Except for a couple of new frameworks that are JS on client and server, which are less than niche at this point, all front-end frameworks work with a PHP back-end.

Fair points. I'd just like to bring your attention back to the point that

"What the application does and how it behaves is all that matters."

I'm hard pressed to find a rebuttal to this, if only for the reason that implementation is the goal. No one will ask you to write an application in Python, "and who cares what it does". The very opposite is true - no one cares what it's written in as long as it's the necessary implementation.

From here I would like to address my "point" that tech stack and framework don't matter. I think that we're actually on the same page here save for some sloppy wording on my part. I concede that on a per-project basis framework and stack matter. They quite clearly do, and I agree with you.

I would still like to argue that they don't matter as much as we like to think. To use the example you gave - PHP - depending on your purpose, your required implementation, PHP is still a powerful and useful tool. It's easier to work with than many other technologies for certain kinds of solutions. At the same time, it might be a better idea to write a solution on Rails for any number of reasons. These are specific to each project. At the end of the day though, you can successfully implement any number of solutions on a great many different stacks with no great change to implementation. I can write an application and use either PHP or Rails for the backend, Angular or Backbone for the frontend framework if I have one, and for the end user there will be ultimately no difference. I can run a PostgreSQL or a MongoDB database and they won't be able to tell the difference. If this is the case then, what makes a good application? It is the people writing the code, architecting, and meeting the challenges in ways that work with the chosen frameworks and technology stacks. It seems to me that the end result, then, is reached by people and not by stacks or frameworks.

Sorry for the long reply. No harm if you don't reply to it.

In my PHP example, I was referring to programming frameworks in general. Not just front-end ones. Many apps have moved away from PHP as developers recognized its weak abstractions which lead to higher maintenance costs. For front-end, you can simply replace PHP with Java pagelets or jQuery soup.

You can write the same web app (from the user point of view) using Python, Ruby or PHP (frontend: jquery, backbone, angular). I do not disagree with this. However, the choice of frameworks is still very important and will determine your speed of development and any future maintenance.

Saying frameworks are irrelevant since they have the potential to all lead to the same outcomes seems a bit much. Any turing complete language can implement program as another turing complete language, but why do we have assembly vs Java vs C vs Python? Higher level programming languages and frameworks are for the human brain. It turns out the the human brain grasps some concepts better than other concepts: we call the more intuitive concepts abstraction.

Of course people are more important than tools but that doesn't mean tools are unimportant. In same cases the tools are VERY important. There are for example things that are easy to do with javascript that are impossible or almost impossible to do with only server side code. If you feel that people are so much more important than tools that the tools almost makes no effect at all please write more about people then. This article wrote almost nothing about getting the right team, getting people to work together, getting the wishes or expectations from the customers etc. It seemed to say that javascript is overused (which it might be) but with a lot of words.

I've worked on projects that have generated hundreds of millions of dollars, all written in shitty PHP code. Was it a big pain for us developers to maintain it? Yes. Did our struggle matter? No.

Well put. Very succinct.

What's missing is the why. Just asserting that all JS-driven sites are broken and should use progressive enhancement doesn't really translate into real benefits.

Using a system that spits out JS and saves me time and effort is a win. Why should I trade that out?

I agree for some users, a properly gracefully degrading system is nice. I like when sites I use are properly URL-based, when I can share links and they just work. But that's rarely a reason I'll switch sites.

So until there's some solid evidence that relying on JS translates into lost money or other serious damage, it's simply not worthwhile and declaring such things broken by fiat accomplishes nothing.

I disagree. Backward compatibility is sometimes necessary to keep and sometimes necessary to break. And the standard of browsing is evolving, with or without such efforts.

If your page is unreadable with NoScript and RequestPolicy turned up to maximum paranoia, your webpage is broken.

Building a broken web page is fine if you choose to build it that way. If 1% of people can't see your site, or 10%, or even 99%, then that's your choice as a developer. If I want my business website built with WebGL, asm.js, HTML5 audio and WebRTC then that's ok. I'm allowed to do that even if you can't render it. As a user you have no implicit right to use a given website if your choice to turn something off or to use a browser that doesn't support something means it won't work for you. To draw an offline analogy, I have no right to build an application if I choose to remove make from my computer.

There are exceptions - users who need special access technology like screen readers should be protected from developers who build in ways that mean they're put at a disadvantage for example - but ordinary people who choose not to use something like js have no such protections. Using a screen reader is usually not a choice.

Reply on news.ycombinator.com