Ask Stack: Should I learn a new programming language? | Hacker News Reader

Ask Stack: Should I learn a new programming language?

arstechnica.com

22 pointsshawndumas11 comments

My problem with this question always is: what do they mean by "learning" a language?

Does it mean you just learn the syntax? Does it mean you can write a simple program that works? Does it mean you can write a simple program that works and exhibits the idioms of the language? Does it mean you can write a medium length program without having to refer to api docs? Does it mean you can write a complicated, robust, production quality system in the language? Does it mean all your friends who have a perplexing problem in the language shoot you an email because you know it inside and out?

All of these take differing amounts of time to accomplish, and all can fall under learning a language. I think I've learned erlang at least 3 times now. When you say you learned a language in a week or two, what does that mean?

Unless you have a particular project in mind, I don't think you can just "learn a new programming language". There has to be some context for it.

Perhaps better advice is to seek out a new project in an area you're not familiar with and use the opportunity to seek out the best tool to solve that problem regardless of the technologies you already know. If you need to build a server process, maybe that would be a good time to look into node.js. If you need or want to design a web application look into Ruby.

A programming language can partially influence the way you think (a form of Whorfianism http://en.wikipedia.org/wiki/Linguistic_relativity), but a PL is a poor substitute for actual thought.

Thinking in terms of any programming language limits the ways you can model problems and solutions. Of course, learning ways of modeling expands your repertoire. I would hesitate to say that "programming languages" are the richest source of such models. Even within them, features like dynamic typing and closures are less significant than logic/constraint programming - or GPGPU languages.

On a separate issue, we can also see programming languages as an instance of exponentially improving technology (Kurzweil). I haven't seen a chart of the productivity improvements of "programming languages" over time, but certainly there's no contention over the dramatic improvements from raw machine code, to assemblers, to compilers etc.

The mechanism is the same as for increasing returns for technologies: standing on the shoulders of giants, and seeing further, especially ways of doing things (the meaning of "technology"). That is, we create the next generation of programming languages using the previous generation (and benefiting from their productivity). e.g. we wrote compilers with assemblers, then with compilers, and then with specialized tools (yacc, lex etc).

At the forefront of technology, it's not always obvious which ways will turn out to be best til afterwards - it's an exploration of the news, which is unknown by definition. Anyone who thinks they know the answers before they start exploring misunderstands the nature of exploration. Before we have definite facts, we have opinions and contention - which is all for the good, if it causes people to look harder rather than argue blindly.

My first admittedly smart ass remark was "If you have to ask, then the answer is 'No'" But that doesn't treat the question as legitimate---which it is. For me the answer has always been 'Yes' but perhaps not for the reasons found in the remarks below the article. I'm more or less addicted to learning computer languages. Part of that is plain curiosity---I just like to see what is in the offering. The rest is the certain knowledge that each new language gives me a new point of view. Sometimes a new solution to a problem; but in so far as language shapes how things are viewed then the addition of additional material is almost mandatory as far as I can see. I think someone mentioned "One a year"---I don't know if I'd set a time constraint on the idea, but I'm usually at work learning a new one at any given moment. I also re-learn old friends in terms of using 'new to me' compilers; Borland->MVC->GCC, that kind of thing. I also believe that anything that keeps the mind active in my chosen field can't help but be a good thing :)

Personal choices aside (everyone has their own) from a professional standpoint you have to consider that even a mildly advanced Ruby programmer gets a bigger paycheck than an experienced PHP programmer does for the simple fact there are less Ruby programmers around and its currently on high demand.

So learning the hottest language can open doors for you, jobs-wise.

Exactly. There are some well-paying niches where a "polyglot programmer" will be required, e.g. C++/Python, Ruby/Javascript... knowing only part of the required combo will likely not get you the job or curtail your remuneration.

A career-minded person could just expand his core competencies in order to nail those niches. Not everyone enjoys hacking LISP in their spare time, but expanding your horizons is the right thing to do if you care about your job prospects. If you write for the embedded market at a low level, pick up something higher in the stack; if you only dabble in high-level languages, try looking under the hood; if you specialize in particular technologies (e.g. C#/.net), look at their complementary stuff (e.g. javascript, powershell, or even Java interop and VC++).

Also, "learning a language" could just be "learn a big framework / toolset". For example, if you know Python, you might want to pick up Django, or a GUI framework like Qt, even if they are outside the boundaries of what you need at work: they tend to have their own programming style and quirks, so they can be key technologies to bring you new opportunities. Java has boatloads of big frameworks, chances are that you don't know all of them. Ruby people tend to live all the time in their Rails ghetto, but there's a world of libraries outside that. And so on and so on.

I like how Python is considered a 'next new thing'. It merely predates the web.

Well no. There may have been a language which was called Python but it seems to have been grown and added to over the years (OO, a faster JIT, more libraries) which suggest to me that the incarnation of Python which predates the web is not the incarnation that became popular.

If you asked me this question, I would assume you are a mediocre programmer. The good ones don't consider it a big investment to learn a new language, because they can do it really fast and because they enjoy it.

Since when did learning a language stop being a big investment? Beyond syntax there are idioms, idiosyncrasies, standard libraries, advanced features, upstream features, and toolchains.

How many people do you know have full understanding of C++ or Haskell? Or even Ruby or Python?

Don't forget debugging and reading error messages. I honestly believe that 90% of learning any language is learning how to solve errors when the compiler or run-time spits them out at you.

Reply on news.ycombinator.com