After a long research I decided to learn Haskell. It is very unlikely that I will ever be paid to use it, but I am pretty sure that after more than 10 years experience in programming, Haskell will greatly benefit my programming skills and thinking.
There was a recent comment by tptacek in another thread which I agree with: you really only need to know a couple of programming languages before you start hitting diminishing returns in terms of what you can implement. We should focus instead on learning new concepts or new problem domains.
I'm trying to revisit statistical modelling and pick up some machine learning basics over the next year.
and what about domains only tangentially related to computer science, like robotics, semiconductors, or digital music?
or what about domains barely related to computer science, but where coding is an invaluable skill, like medicine/genetics, manufacturing, or aerospace?
consider how many programming languages your average person knows (zero). now consider how many you know. is it objectively worth your time optimizing around that particular dimension within your skills, or are you perhaps suffering from a latent bias you have from the fact that you are a relative expert in programming?
would you rather have a carpenter who knows how to use a dozen hammers, or a wide variety of tools? is haskell a new tool, or is it just another hammer for you? with coursera you have no excuse to avoid stepping outside your comfort zone!
I'm heads down on golang right now and will continue to be so. I never quite acquired C - wrote plenty of code in it, but it was always a struggle. Go's just above the threshold where my mind will take it wholescale (ala Python) and it gives me the ability to write simple code and generate cross-platform binaries from it.
One app: web, windows, mac, linux. I know that can be done with other tools too, but this is the one I've chosen. I like SQL too: old guys like weird tools!
After programming for 30 years (and upwards of 20 languages) I've come to believe that you really only "need" to learn 4 languages to complete your Jedi programming skills and become a master. You need one each from the following categories:
"Machine" language - this is anything turing complete that's close to the hardware, ie that forces you to think about and understand the underlying circuitry. Candidates: Assembler, C.
Static/Imperative "Object" language - put this one off for as long as you can, for it is easy but not simple, and will ensnare your mind with the belief that you are special because you are among the elite priesthood of programmers who can grasp levels of complexity mere mortals dare not even dream of. Down this path lies years of frustration, all in service of the darkside. Chances are that this category will be forced upon you first, by naive, shortsighted fools who little realize the evil they are spreading. Learn only what you need to disable the tractor beam and escape with the Deathstar layout, so that when the time is right a well planned rebel assault may remove this abomination from the universe.
Candidates: there are no candidates here, as that implies choice. One of these languages will be heaved upon you whether you like it or not - C++, Java or Visual Basic (if the darkside is particularly strong in you).
Dynamic "scripting" language - this is for that category of problems where most of the needed abstractions are well understood and well tested. Using them should involve as few words as possible.
And finally, what should be every programmer's last language,
A "language" language - these are usually known as Lisps, but I think both Alan Kay's original bootstrapping Smalltalk and Forth are acceptable here. The key feature is having a small core of concepts that can be infinitely composed into higher abstraction commands that are treated no differently from the compiler's perspective. Once you have this power, all of the interesting behaviours of languages like Haskell, Erlang, Prolog, etc, are available to you.
Candidates: Scheme/Racket, Common Lisp, Clojure, Smalltalk, Lisp.
Having learned perl, ruby and python (these are all incredibly similar), and then Clojure, I've found that I really like the functional paradigm. It helped drag my attention back to thinking about data structures and functions to act on that data a lot more than a ton of bad OOP examples ever have. I'm actually excited to program now instead of feeling like, "Okay, this mess of classes I'm making is going to pay off some day..."
One negative/positive is that I now kind of recoil in distaste when I find OOP-y code that's seemingly done for the sake of jamming code into a class instead of using generic functions and data types. It's funny that there's a large swath of not so useful monad tutorials, because there's an absolute mountain of examples of classes/objects that don't make any sense at all. I didn't mean to make a tiny rant on OOP. Sorry about that.
As someone who has been using a lot of dynamic languages, I'm really interested to find out how a static type system can help me program better software.
If you're learning Haskell, be prepared to feel like a beginner for a while.
I had great learning it. I usually pick up languages fairly quickly, but Haskell eluded me for more months than I'd care to admit. There is a great community of very smart and friendly people who helped me on IRC, and a couple of pretty good books -- http://learnyouahaskell.com/ and http://www.cs.nott.ac.uk/~gmh/book.html.
I enjoyed the experience, and I certainly feel like a better programmer for it. But damn I felt daft for a few months!
I am looking forward to that experience :) Besides the technical reasons (nice syntax, pure fp, strong typing, ...) I chose Haskell to learn functional programming because it has a very good community and documentation. It's like the entire Haskell ecosystem's been built to teach FP. Actually that was kind of the intention when Haskell was born as far as I know.
I spent the last year doing lisp as much as I could, and I've become a right convert. I've got some C and python down already.
This year, I am currently torn between:
-Haskell (to go full blown functional crazy)
-Smalltalk (to go full blown object crazy)
-Javascript (to finally write some stuff for the web/browsers)
- Java (to sell out and lose the will to live?)
Any advice is more than welcome. I'm probably looking for the "programming language that changes the way you think" aspect, since I've already got a paying job...
PiL will present examples that will change the way you think in several different ways, which you will then be able to leverage in any other programming language (most of the time, depending on how restrictive the language is).
You have Lisp knowledge, so maybe Haskell is not the best candidate for "a new way you think". Given your background I would say you probably would enjoy Scala more than Java (targeting the same JVM). I am not sure if one can learn enough from Smalltalk to justify the effort it takes to learn an in-practice useless language. If you don't speak JavaScript yet, that's a good candidate, especially if you use your FP experience in it and not just go the OO way most people do.
Can someone make recommendations for me? Here's what I already know and what I've built with it:
- Java. My first real programming language, learned in college, used for a year and a half in enterprise work and Android, got pretty good at it, but I learned
- Objective-C for making iOS apps and I never want to write in Java again. I'm also learning C just to understand Objective-C more but I have no plans building anything with C.
Basically, my current interests/goals for learning a new language are:
- Develop a new way of thinking.
- Learn some great engineering practices.
- Build products (software) that solve real-world problems. I'm not likely to learn a new language if I can't build anything useful with it, that can potentially sell and make it as a consumer startup.
- Become full-stack. I can already easily build an iOS app (I have no interest in building responsive websites) but I find myself still having to rely on something like Parse. Which isn't so bad though because it saves me so much time.
I guess that narrows my choices down to: Ruby, Python, Go, or something else, but I'm not in a position to identify which poison is best-suited to me. I just want one that I can be really good at and meet most if not all of my goals. Any thoughts?
Very pragmatic approach. C# if you are ok with the Microsoft stack, or F# if you are a bit more advantageous and serious about the "new way of thinking". Python is also a great investment to solve real-world problems. And don't forget JavaScript. It is not just a browser language any more: there is node.js for the server side and it is the common language of most mobile frameworks that are multi-platform targeted.
You cannot go wrong with any of these. So have a look at code samples and spend some time browsing the web-sites about all of them then choose the one that appeals the most to you.
There are so many languages I would like to learn :
- Ada : everybody raves about it but nobody uses it ;-)
- Python : for everyday hacks, it's portable, simple and it's not Java ;-)
- Assembly : it's the root of everything
- C : my mind is overflowing with Arduino projects
- C++ : I'd love to create ISX extensions and make private bots for some MMO
- Prolog : just because
I will probably settle for Ada if John Barnes publish its book at the beginning of 2014.
Python - I'm doing far too much stuff in PHP just because that's what I know, seems to be fairly simple to learn as well, and quick which should help in the $dayjob efficiency.
I learned Dart and TypeScript this year and both were fairly trivial to learn and use. The learning process improved my JavaScript skills because I had to think about JavaScript in new ways and got me ready for ES6 (For example, the discussions about iterators in Dart and ES6 was interesting.) Next up is ClojureScript, if I can get over the syntax.
I've started to play with Nimrod and I will definitively continue in 2014. Macros are not always easy to use but they are clean and powerful (I'm mostly a C programmer ...)
[+] [-] yen223|12 years ago|reply
I'm trying to revisit statistical modelling and pick up some machine learning basics over the next year.
[+] [-] gfodor|12 years ago|reply
or what about domains barely related to computer science, but where coding is an invaluable skill, like medicine/genetics, manufacturing, or aerospace?
consider how many programming languages your average person knows (zero). now consider how many you know. is it objectively worth your time optimizing around that particular dimension within your skills, or are you perhaps suffering from a latent bias you have from the fact that you are a relative expert in programming?
would you rather have a carpenter who knows how to use a dozen hammers, or a wide variety of tools? is haskell a new tool, or is it just another hammer for you? with coursera you have no excuse to avoid stepping outside your comfort zone!
[+] [-] leashless|12 years ago|reply
One app: web, windows, mac, linux. I know that can be done with other tools too, but this is the one I've chosen. I like SQL too: old guys like weird tools!
[+] [-] zenshade|12 years ago|reply
"Machine" language - this is anything turing complete that's close to the hardware, ie that forces you to think about and understand the underlying circuitry. Candidates: Assembler, C.
Static/Imperative "Object" language - put this one off for as long as you can, for it is easy but not simple, and will ensnare your mind with the belief that you are special because you are among the elite priesthood of programmers who can grasp levels of complexity mere mortals dare not even dream of. Down this path lies years of frustration, all in service of the darkside. Chances are that this category will be forced upon you first, by naive, shortsighted fools who little realize the evil they are spreading. Learn only what you need to disable the tractor beam and escape with the Deathstar layout, so that when the time is right a well planned rebel assault may remove this abomination from the universe.
Candidates: there are no candidates here, as that implies choice. One of these languages will be heaved upon you whether you like it or not - C++, Java or Visual Basic (if the darkside is particularly strong in you).
Dynamic "scripting" language - this is for that category of problems where most of the needed abstractions are well understood and well tested. Using them should involve as few words as possible.
Candidates: Ruby, Python, Perl, Lua, bash (or any bourne shell derivative).
And finally, what should be every programmer's last language,
A "language" language - these are usually known as Lisps, but I think both Alan Kay's original bootstrapping Smalltalk and Forth are acceptable here. The key feature is having a small core of concepts that can be infinitely composed into higher abstraction commands that are treated no differently from the compiler's perspective. Once you have this power, all of the interesting behaviours of languages like Haskell, Erlang, Prolog, etc, are available to you.
Candidates: Scheme/Racket, Common Lisp, Clojure, Smalltalk, Lisp.
[+] [-] Myk267|12 years ago|reply
Having learned perl, ruby and python (these are all incredibly similar), and then Clojure, I've found that I really like the functional paradigm. It helped drag my attention back to thinking about data structures and functions to act on that data a lot more than a ton of bad OOP examples ever have. I'm actually excited to program now instead of feeling like, "Okay, this mess of classes I'm making is going to pay off some day..."
One negative/positive is that I now kind of recoil in distaste when I find OOP-y code that's seemingly done for the sake of jamming code into a class instead of using generic functions and data types. It's funny that there's a large swath of not so useful monad tutorials, because there's an absolute mountain of examples of classes/objects that don't make any sense at all. I didn't mean to make a tiny rant on OOP. Sorry about that.
As someone who has been using a lot of dynamic languages, I'm really interested to find out how a static type system can help me program better software.
[+] [-] ciderpunx|12 years ago|reply
I had great learning it. I usually pick up languages fairly quickly, but Haskell eluded me for more months than I'd care to admit. There is a great community of very smart and friendly people who helped me on IRC, and a couple of pretty good books -- http://learnyouahaskell.com/ and http://www.cs.nott.ac.uk/~gmh/book.html.
I enjoyed the experience, and I certainly feel like a better programmer for it. But damn I felt daft for a few months!
[+] [-] peterfoldi|12 years ago|reply
[+] [-] ACow_Adonis|12 years ago|reply
This year, I am currently torn between:
-Haskell (to go full blown functional crazy) -Smalltalk (to go full blown object crazy) -Javascript (to finally write some stuff for the web/browsers) - Java (to sell out and lose the will to live?)
Any advice is more than welcome. I'm probably looking for the "programming language that changes the way you think" aspect, since I've already got a paying job...
[+] [-] spb|12 years ago|reply
PiL will present examples that will change the way you think in several different ways, which you will then be able to leverage in any other programming language (most of the time, depending on how restrictive the language is).
[+] [-] peterfoldi|12 years ago|reply
[+] [-] dyscrete|12 years ago|reply
[+] [-] mattquiros|12 years ago|reply
- Java. My first real programming language, learned in college, used for a year and a half in enterprise work and Android, got pretty good at it, but I learned
- Objective-C for making iOS apps and I never want to write in Java again. I'm also learning C just to understand Objective-C more but I have no plans building anything with C.
Basically, my current interests/goals for learning a new language are:
- Develop a new way of thinking.
- Learn some great engineering practices.
- Build products (software) that solve real-world problems. I'm not likely to learn a new language if I can't build anything useful with it, that can potentially sell and make it as a consumer startup.
- Become full-stack. I can already easily build an iOS app (I have no interest in building responsive websites) but I find myself still having to rely on something like Parse. Which isn't so bad though because it saves me so much time.
I guess that narrows my choices down to: Ruby, Python, Go, or something else, but I'm not in a position to identify which poison is best-suited to me. I just want one that I can be really good at and meet most if not all of my goals. Any thoughts?
[+] [-] peterfoldi|12 years ago|reply
You cannot go wrong with any of these. So have a look at code samples and spend some time browsing the web-sites about all of them then choose the one that appeals the most to you.
[+] [-] jmnicolas|12 years ago|reply
- Ada : everybody raves about it but nobody uses it ;-) - Python : for everyday hacks, it's portable, simple and it's not Java ;-) - Assembly : it's the root of everything - C : my mind is overflowing with Arduino projects - C++ : I'd love to create ISX extensions and make private bots for some MMO - Prolog : just because
I will probably settle for Ada if John Barnes publish its book at the beginning of 2014.
[+] [-] aldanor|12 years ago|reply
[+] [-] tsahyt|12 years ago|reply
[+] [-] iSloth|12 years ago|reply
[+] [-] deadfall|12 years ago|reply
[+] [-] michaelwww|12 years ago|reply
[+] [-] hkarthik|12 years ago|reply
I think there's a lot of interesting ideas around high availability and fault tolerance to explore in the Erlang world, and I'd like to tap into that.
[+] [-] cju|12 years ago|reply
[+] [-] phektus|12 years ago|reply
[+] [-] spb|12 years ago|reply
[+] [-] tdfx|12 years ago|reply
[+] [-] nnoitra|12 years ago|reply
[+] [-] Pishky|12 years ago|reply
[+] [-] anwaar|12 years ago|reply