Hacker News Reader

Comment by polysics | Hacker News Reader

What do you reckon could not be legacy code in 15 years? C family, Java, maybe Erlang?

I expect the JVM to still be going; at least one of Clojure or Scala will probably continue growing for a long time, and there's room for both. It's also possible another Algol-esque language that isn't Java, but is syntactically closer to Java, making people more comfortable, will arise. At some point one of them may even become the "primary" JVM language, with Java hanging on as legacy, but due to the fact Java just compiles to the JVM, there won't be any reason to get rid of it, as the maintenance effort is virtually 0. I expect C# to probably still be going. I'm not sure anything can actually kill C, even though something really should.

C++ I can see going either way; it is a freakishly complicated language (and getting more complicated with every standard that comes out) that IMHO was primarily kept aloft by being the "default" language in Windows for a very long time. I expect Rust will be most of the way through eating it alive in 15 years. Or something like Rust, but Rust is looking pretty solid. And if you want a multithreaded systems language... and you will... I suspect in practice it's going to be an order of magnitude easier to program in than C++ in another three or four years. Maybe Rust will even be eating C. (About damn time!) Someone will probably be writing an OS kernel in Rust. Possibly one that is driver-compatible with Linux, or even just straight-up a Linux-compatible kernel.

If Google doesn't drop the ball, Go will probably be mature and still going, and probably the primary reason why Ruby and friends are considered legacy code. Go even today very nearly replaces the scripting languages in flexibility (though definitely not quite), but brings back stronger typing and better performance. A few more tweaks to it over the next 5 years and it'll be hard to see why I'd start a new project in a "scripting" language. (Part of the reason I think this is that I observe that in the past 15 years, no language has penetrated into the A-list without some sort of major corporate backer. While there's plenty of other possibilities from a technical point of view, and while Google is not very aggressively pushing Go, I still think it's more support than any other language has or is likely to have in the near future. Even if they just passively continue to do little more than employ the authors, this is likely to continue floating up the ranks.)

And I still sort of expect something to come up that is not Haskell, but is closer to Haskell than anything currently existing. Or possible Haskell simply keeps plugging along and manages to become a very solid B-list language. I doubt it'll ever break into the A-list "nobody got fired for choosing C/C++/Java/C#" level of popularity, but it might settle firmly into where "it's pretty hard to get fired for choosing Python/Ruby/PHP" is today. (It's brushing B-list status now, but it's got a ways to go before I'd call it solid B-list.) I also think there's a good chance it will acquire a reputation as being the tool of choice for tackling truly complicated systems. (Though Rust will eat into the bottom end of that a bit too, they're currently retreating from some of the tools that Haskell has that may allow it to acquire this reputation. I think Haskell will still retain a strong advantage here.)

And finally, some sort of wildcard. 15 years ago I doubt anybody would have predicted Haskell being where it is now, even Haskell users (if you can even call it the same l. New languages shall continue to bubble and froth, but at least one and probably two will either emerge from the froth and become a B-list language. Probably at least one of them will be a aggressively "practical" language that will be a reaction to the success of "excessively-academic" Clojure, Scala, Rust, and the Haskell-like I hypothesize above, and follow a hype curve very similar to Ruby or Node.

It occurs to me I'm looking more like 10 years into the future here than 15. Oh well. Close enough. I'm also well aware I'm just talking out my ass; I completely expect to look back and laugh at this in 10 to 15 years. Oh well again.

I like Go, but in my opinion it is not an appropriate language for building web applications. It makes for great sub-components of a larger stack which are focused and do relatively low-level tasks (networking, file serving, caching, etc.). It is also great for scripting where you need easy out-of-the-box concurrency.

This times a million. I think Go is awesome for a little bit of systems programming but I'd pull my hair out trying to make a robust Web application with it. I feel the same way about Node.js. I wouldn't use either of these for a public facing HTTP application but if I needed some hyper performant internal system using RPCs then I would use them.

Ruby's DSLs for writing things that Web application's need (serving HTML) are bar none the best. If I'm writing something that is serving HTML, I'm using Ruby all day, every day.

Have you given martini a try?https://github.com/codegangsta/martini

For a Google-backed language to replace Rails, Golang has less of a shot IMHO. Mainly because the popularity of Rails stemmed initially from TextMate in the original screencasts. It'd be silly to say that's all that mattered so I'm going to merely strongly imply that without additional books or cookbooks, screen casts, etc. it's hard to see Golang taking off in a community that isn't already invested in learning it. Those looking for more stability will pick a JVM-based language, which includes JRuby, and call it a day.

If I had to then pick a replacement, I would argue for Dart. Dartlang has a much stronger set of novice documentation, nodejs-like magic of running in client and server, but advanced Google mojo like support for SIMD. All Dart really needs is Chrome support and a decent "Rails" clone to popularize server-side development. We're a few years off, but Dart could be quite exciting. Or it could end up running on the JVM like everything else :p

"For a Google-backed language to replace Rails"

It's not "replacing Rails", it's replacing new projects in what are today scripting languages (Perl, Python, Ruby, PHP, etc.).

I see fewer people talking about Dart than I see doing things with Go. That said, I did finally beat Google Trends into producing [1], which is interesting. It was hard to convince Trends to show Dart as growing as quickly as Golang, though, so there may be a selection effect (of the several graphs I produced with Dart well behind Golang, this is the only one that shows them even tied), and it still shows Dart as tied with Go, and Go's growth being more steady. I'm still not sure Dart is growing like Go is.

I suppose I should point out that the predictions were my best guesses about what will happen, and do not simply reflect what I "want" to happen.

[1]: http://www.google.com/trends/explore#cat=0-5-31&q=dart%2C%20...

Go's been out 4 years while Dart just hit 1.0 and requires a runtime as it's a bit less portable. Once Dart lands in Chrome, I can easily imagine it becoming the first choice of developers who prefer to develop code in their browser. A niche market now, but interesting to contemplate.

I'm a bit more flexible when it comes to trying to predict the future than my posts indicate. I'm just offering an alternative hypothesis: Dart has many more screencasts and supported libraries from Google, including Angular.dart than Golang ever had, and it's only just hit 1.0. Golang has been fantastic for me to develop with, particularly on Google AppEngine and it's true Dart hasn't hit server-side popularity yet. But if we're predicting the future, based on JS and node.is alone, I'd suggest Dart has plenty of room to gain popularity quickly.

Oh, I see, that makes sense.

For me, I was making broad predictions. No matter how exciting the web client dev is, it is and will remain still a niche, and Dart faces a huge challenge in the next 5 years as asm.js will relatively suddenly introduce it to competition from every other mainstream language in existence. Dart probably won't "die" for a while, but I'm not sure it won't end up "just another" one of the enormous pile of languages that it will be possible to do client-side programming in.

I've learned to not bet against the ingenuity of Google engineers, particularly those building web browsers and development tools. :) Given that they've added very basic support for editing to Developer Tools in Chrome already, once Dart lands in Chrome, I could see a push to make Chrome the editor-of-choice for Dart, if not web-based editors in general. And it's true, there's more to development than type-and-run interactive shells, but that sure makes web development easier, doesn't it? asm.js benefits Dart as well, since they can target the same approaches for their JS target code gen.

The real question is when will Google add development alternatives to Android? Chrome's nice, but it takes a mobile platform to sustain an ecosystem of client-side developers these days. Perhaps Chrome-on-Android will eventually become the new PhoneGap?

AFAIK Dart still does not have generators. I want to like Dart but if puts me in the same boat as Javascript does on the server, then NO.

> Someone will probably be writing an OS kernel in Rust.

There are two or three of these, at least at the stage where kmain is Rust code. I'd be really excited to see this develop further.

I just don't have time for ever project I'd like to do :/

By that prediction, I did not just mean that someone wrote something in Rust that can boot and do kernel-y things... I mean that there will be a kernel project in progress that is not merely "intended to be serious competition someday", but something that you can seriously use for real, production projects as your computer's (or server's, or cell phone's, or something's) primary kernel/hypervisor/the thing actually running on the hardware. Indeed one of those projects may someday even become the kernel I am predicting, but I'd be very surprised if they are something anybody would seriously make their primary runtime today.

In other words, this is an extreme statement of confidence on my part that Rust is very likely to succeed in their goals of being a true system language replacement. (Contrast this with Go, which initially started out saying it was going to be "system" language, but they've clarified they mean something more like a serious application language rather than something you'd write an OS in. "Nobody" is going to write a kernel in Go.)

Absolutely, I agree. I was just trying to say that hopefully, there's some intrepid coder out there trying to do this, and that we have a few people who have started. We'll see where they are in a few years. :)

You seem down on both C and C++, but there will always be demand for languages that allow you to control memory.

Rust looks promising, but it's way too early to tell if it can really replace C/C++.

I think what jerf means is that, being mostly used as a scripting and web language, it's easy to create legacy code in it. Remember ColdFusion? PHP 3?

However, more to your point, there's legacy C and Java code, but that's because those languages have evolved heavily in their time, so late 90s Java looks little like modern Java, and techniques used in C in the 80s would get you laughed at today.

Hell, even in my main language of work, C#, the difference between most code just 4 or 5 years ago and "modern" C# is vast.

That's pretty fascinating...I studied C in college and didn't hate it (in fact, I'd like to go back to C some day)...but my perception since has been that not much has changed because not much needs to change. I'd love to see a paper or article looking at innovations or change in practices in these older languages.

Not the OP, but I would expect to see still widely used:

* Python - As a programming language for scientists and researchers, replacing Matlab and kin.

* Go or Rust - A well-designed, modern language, easier to use than C++ but almost as fast, for enterprise applications or video games. Not sure about Google, but I hope Mozilla will still be around in 10 years.

* Lua? - Not sure about this one... but it fills a niche -- extremely compact scripting language.

* C - still used in system-level programming

* C# - C# is replacing Java as the OO language of choice for big enterprise systems.

C# - C# is replacing Java as the OO language of choice for big enterprise systems.

I'm skeptical of this one. C# requires Windows Server or putting up with Mono's shit. The former sucks; the latter isn't great either. I'd bet that iterations on Java, as well as other JVM languages, will probably retain a significant edge long-term.

In my experience (admittedly limited to only 3 specific instances), large orgs. have absolutely no problem with using Windows Server (often alongside Linux servers as well). B2B technology is remarkably Windows-centric even now, and I expect it will only become more so as time passes and Microsoft improves its server software and the .NET ecosystem, which is still pretty immature compared to the JVM.

"C# - C# is replacing Java as the OO language of choice for big enterprise systems."

C# is held back by windows. I know there is mono, but enterprises aren't going to bet on mono. So I have to disagree with that.

Rumor goes that MS will likely make C# cross platform or open source entire .NET runtime, so that they can run Dynamics on Linux as well. In my opinion, if they did this, I would expect C# to get significant growth, and eclipsing Java.

C# is held back by windows

"The enterprise" (for the most part) isn't held back by Windows. That's just reality.

> "The enterprise" (for the most part) isn't held back by Windows.

Or, rather, much of "the enterprise" is committed to being held back by Windows, so a language being tied to windows doesn't hinder enterprise adoption.

Except Windows Server works fine.

Why group rust and go? Rust allows control over memory and go does not, which seems like a huge difference.

I'm not so sure about Lua. I'm under the impression that it's primarily used as a scripting language for games and things like that.

The language is small, it's very fast as scripting languages go, and I've heard it's easy to integrate it with C and other languages. On the other hand, it has kind of weird syntax, breaks a lot of conventions (i.e. arrays are 1 indexed), and it's not a massively well known language.

JavaScript possesses a lot of the same benefits; it's a pretty simple language, boasts very impressive performance for a scripting language, and isn't very difficult to integrate with code written in other languages. As compared to Lua's downsides, it's a bit of a mixed bag. The syntax is immediately familiar to anyone whose used an Algol derivative, the language has some unpleasant quirks, but lots of developers are accustomed to them due to the language's substantial popularity.

I imagine Lua will still be a popular choice in 5 years, but I think it's going to lose some marketshare to JavaScript.

What makes you think Google won't be around. If they weren't, I'd put Mozilla on the chopping block as well.

[1]http://thenextweb.com/insider/2013/11/21/mozillas-reliance-g...

"* Go or Rust - A well-designed, modern language, easier to use than C++ but almost as fast, for enterprise applications or video games. Not sure about Google, but I hope Mozilla will still be around in 10 years."

Unfortunately I imagine that Google has enough money to be around at least 20-30 years.

Unfortunately? Yeah Google deserves to die, self driving cars, actual robots, Google glass, android, search, map reduce etc. Who needs this useless company that doesn't contribute anything to the world, right?

Reply on news.ycombinator.com