Ask HN: How do experienced programmers learn new languages? | Hacker News Reader

Ask HN: How do experienced programmers learn new languages?

32 pointsmercsilv19 comments

I imagine this is different once you have a few languages under your belt - so I ask you, how do you go about this?

This was taught to me in college a decade.

"Programming language is just a syntax, the core logic of getting things done is already available in your brain."

Here is what i'd do

A) Remember what you already learnt from your favorite language

  1] Syntax

  2] Libraries

  3] Reference Documentation

  4] Organizing your code

  5] Building / Running Programs

  6] Deploying code to farms of servers (if you are interested in sysops, explore this as well)

B) Once you get a hang of basic syntax, start with Examples and keep making progress.

C) Pick a simple task, from what you have already solved before. Program it in the new language by expanding your skills towards 2-5 points and you are done !

"Programming language is just a syntax, the core logic of getting things done is already available in your brain."

Is this actually true, though, when jumping between programming paradigms?

My only experience learning a new language that use a completly different paradigm is Prolog. The syntax and builtins make it hard to write something else that what it is meant for.

Do you mean learning or mastering a language. It is kind of different. Looking for "ideomatic python" will help getting pythonic Python code done.

Not going against the spirit of the language helps.

Well, even just using a language (without mastering it) can be a lot more difficult than simply learning the syntax.

For example, many years ago, before I really got functional programming, I tried to learn ML and while I figured the syntax out very quickly, solving problems in a functional style was a lot more difficult and I found it hard to use the language. Learning to use Python or Ruby or Java or C#, albeit not idiomatically, is a hell of a lot easier.

Of course, knowing a language is a lot more than just learning enough to use it, though you certainly don't have to master a language to know it (and many people will never completely master their languages, despite knowing them quite well and using them effectively - for example, I'd wager few people have truly mastered C++).

No, it doesn't really hold up. To use an example of where I recently had to change the way I thought of things: you might have, say, an intuitive understanding of the logic to make a series of dependent network requests, any of which could fail, but you'll have to reevaluate that when you try a functional language with monadic futures.

Many languages have a few "killer" features/libs/uses. I often start with a project that uses one of them. Eg:

ruby => rails

python => nltk,django,opencv,scipy

c++ => opengl,opencv

java => opencv

js => node,angular,ember

clojure => overtone

That way you learn the language and some specific good use cases of it.

I think it depends on how different the new language is. If you have a basis in C-style languages, it is often not to hard to move between them syntactically. Nuances might take a bit more time -- for instance, going from C/Java to C#, lambdas, LINQ, and other commonly used components took a bit of time.

When I first took a gander at Erlang, I poked at it, got a bit bored because reading the language wasn't a bit harder. I took some time off, revisited a couple of books and looked at some code and it made more sense.

But, overwhelmingly the way to learn is to find a small project that you are interested in implementing and star there.

There really are two parts to learning a language -- the syntax and the ecosystem/standard libraries. Often, becoming accustomed to the syntax is the easy part, understanding the ecosystem can take more time.

I agree 100%. One of my favorite ways to learn a new language is by doing problems from https://projecteuler.net . The mathematical basis of Project Euler seems to map into core language concepts better than many other "getting started" exercises.

I don't know about other people, but the only way I really learn a new language is by dedicating myself to build a project using it, no matter the size, but it has to be something that you have a clear vision of what you want to build (so you struggle at the hard parts, instead of just changing the course of the project). This will make you search for ways to "say" what you need in that language. It's analogous to going to another country to learn a foreign language.

I fiddle with some tutorials, get frustrated with readability / pacing / coverage of core concepts, then knock out a project or three with the help of Google, StackOverflow, and the docs for the language and any libraries that seemed relevant at the time.

Across, I don't know, 20-25 languages at this point? Although I only actively use 3-5, depending on how you want to define actively.

Read through the documentation. Write some simple code to learn toolbox idioms and patterns. Read through the docs again. Look at some simple code. Build something non-trivial. Then read through the docs again and you're ready to go.

Learning something new in your field of study is not different, it is just easier because you have some context.

If you have a background with a Computer Science degree, it makes it easier. One of the later courses in the program I had was assembly programming. Everyone said this was the class that weeded out people. I found it easy because there were only so many ways to do things and you really begin to understand how computers work at the lowest software level possible.

Once you have a basis of data structures, object-oriented programming, and an understanding of libraries in one area (say web programming with Java Servlets) then learning another language is just a matter of building something and reading the API docs as you go.

Try re-writing something not trivial but not overly huge to learn the differences between old and new. After a while they all look the same, regardless of how trendy or hip something might be.

There probably is something you have done lots of times, mildy complex, but you know how to write it in the languages you have done before. With the new language you take that example and try to see how it would apply to the new language. You know what you want to do and what to expect for output so most of your brain is occupied by implementing in in the new language.

I've read folks who have some game such as tetris or some other as an example.

Also since I know various language commands and methods, I'll read through the language reference from ! to z and see how it compares to what I know and probably take notes along the way.

Do one or two tutorials, then jump straight into a moderately-sized project. Google profusely but also don't be afraid of hitting the language's manual.

Perhaps most important is to not settle for code that you know is not ideal. If you recognize that a piece of code you've written isn't good, then improve it. If it's still not idea, improve it again. Keep iterating on it. This will necessarily involve focused learning. In fact I think this is the key to picking up a language quickly. But it does require that you not be mentally lazy.

Think of something you want to write in that language; this is the only thing that really works for me.

Tutorials and canned examples are ok, but if it's not something I actually want then I'm not really learning much.

And if you can't do anything useful in the language you want to learn, then ask yourself why you want to learn it?

Or take something relatively simple, and code it in whatever language. I did this a while back, and wrote a port scanner in c, java, perl, and python.

I keep a list of side project i want to build. When i begin a new project and the new language/technology i think might be a fit, i'll build something in it. THis forces me to think outside the usual hello world tutorial and building another to do list. And the problems I face to build my custom project forces me to think and get into what i'm learning

Do a tutorial or two, then just start messing around and building. I find that if I try to build something useful, I end up learning the code step by step as I understand how it works and the features that are available

I was once told that the best way to cement your knowledge of a new language is to build a chess game in it.

this a starting point: http://learnxinyminutes.com/

might help: http://rosettacode.org/wiki/Rosetta_Code

reading about the language or read code. Writing code...

Reply on news.ycombinator.com