Learning new (programming) languages

Topics/tags: The joy of code, rambly, long

As I think I mentioned in my most recent musing on my sabbatical, one of my goals for this sabbatical is to explore a few new languages. What languages? Right now, I’m inclined to look at Clojure, Kotlin, and Rust. I may also look at the incorrectly-named C Sharp [1] and a few of the new dialects of Smalltalk, such as Pharo [2]. In addition, it’s probably time to continue my deeper dive into Racket as my knowledge tends to be of the core Scheme aspects of Racket, rather than its newer features. Some colleagues will probably tell me that I should re-learn Haskell, which I last used seriously in about 1995 [3]. Since there have been significant updates since then, it’s likely a very different language.

Why do I want to learn new programming languages, other than to increase the number of programming languages I can put on my imaginary résumé [4]? As I’ve said, while the natural language you speak may not affect how you think [5], the programming language(s) you use can significantly affect how you approach problem solving. You’d think that after thirty or so years teach CS, there wouldn’t be more for me to learn. You’d be wrong. For example, learning Ruby just a few years ago helped me reconceptualize some issues in object-oriented programming [6].

So, why this particular set of languages? Let’s start with Clojure.

Clojure

As part of my not-quite-successful pledge to read something new each day, I read an article by Robert C. Martin entitled Why Clojure in his The Clean Code Blog.

This ’blog post was referenced on the Racket Users mailing list. I appreciate reading yet another experienced programmer who eventually realizes that the Lisp family of languages are the best way to go for most projects. Martin focuses on Clojure, a Lisp-like language. From what I can see from his posting, Clojure has some interesting variations of the standard Lisp syntax, including a hash for something like lambda and implicit arguments. Here’s how he writes a program to print the squares of the first twenty-five integers.

(println (take 25 (map #(* % %) (range))))

The implicit parameter (%) is an interesting choice, as is the implicit lazy evaluation [7] or at least a tightly linked stream model [8]. In any case, the post suggests that I should probably spend some time learning Clojure. As I mentioned in the notes on my sabbatical, I may even consider what 151 would look like if taught in Clojure [9].

I also appreciate two particular paragraphs from the post.

Smalltalk was small and elegant and beautiful. It spawned the Design Patterns revolution. It spawned the Refactoring revolution. It spawned the TDD revolution. It helped to spawn the Agile revolution. Smalltalk was a language of tremendous impact.

Finally, Lisp is functional. And the future is looking very functional to me.

I’ve been predicting success for functional programming for more than twenty years, so it’s good to see someone relatively famous say it, too. And the paragraph on Smalltalk reminds me that I need to learn more about Smalltalk.

Other than supporting my world view, what is it about Martin’s post that makes me want to learn more about Clojure? In part, it’s that Clojure seems to have chosen a different approach to Lisp-like languages. In part, it’s that Clojure can compile to both JVM [11] and JavaScript [12]. That means that Clojure programs can easily run on a variety of platforms, including on the Web. In each case, it means that one can take advantage of the large set of libraries associated with each platform.

Kotlin

I know almost nothing about Kotlin. But one of our recent alums, Amanda Hinchman-Dominguez, is involved in the Kotlin community and speaks highly of it. Like Clojure, Kotlin compiles to JVM code [15], giving it the advantages of a widely adopted platform and a large collection of libraries. Youngest Son tells me that Kotlin is becoming the language of choice for Android development [16].

Kotlin comes from the IntelliJ community or, more precisely, from part of the JetBrains corporation, makers of IntelliJ. IntelliJ is one of the most popular IDEs [19] for Java. We don’t tend to use IntelliJ at Grinnell because it’s not open source. However, given our problematic experiences with Eclipse over the past few years, there’s a chance that we may consider it again [20].

At first glance, Kotlin looks like yet another language with C-like syntax and modern object-oriented and functional programming features—such as classes, lambdas, streams—as well as a bunch of useful syntactic sugar [21]. And it seems that the Kotlin folks take functional programming seriously; I guess I should not be surprised that at least one Grinnell alum has a role in the Kotlin community.

Are there other reasons to learn Kotlin? I can see some advantages to knowing a modern language that is growing in popularity. And a lot of programmers seem to really like Kotlin [23].

I also see that the folks at JetBrains have an educational platform. That may also be worth checking out [25].

Other languages

What about Rust, Smalltalk, C#, Racket, Haskell, and the rest I mentioned? I’ll leave discussion of those languages to another musing or musings. Perhaps I’ll muse about those as I learn them, rather than in advance.

Learning languages

I find that I approach learning languages in a variety of ways. I try tutorials, but I often find them frustrating. Why are they frustrating? At times, it’s that they are too slow for an experienced programmer. At times, it’s because I consider the approach they take wrong [27]. Nonetheless, tutorials can be a good way to start.

When I know a bit about a language, I often write meaningful, or at least somewhat meaningful, programs. I regularly program to meet my needs or for fun. Writing or rewriting a program in a new language can help me learn more about that language. For example, I deepened my knowledge of Ruby by rewriting my grading software in Ruby [28]. That rewrite also helped turn a Perl hack into something a bit more elegant. I deepened my knowledge of Liquid and Jekyll by working on my course Web sites [29].

I often find it useful to read books about new languages. Reading books gets me away from the computer [30] and can encourage me to think more broadly about subjects. However, I’m primarily a hands on learner, so much of my reading inevitably pushes me toward the keyboard. Nonetheless, books are useful. While I think I learn better from printed books, I’m finding it increasingly useful to read PDFs on my iPad, since I can more easily mark them up and copy key portions to a separate set of notes. I also appreciate the free access to Safari I get along with my membership in ACM [31,32]. I have to see what note-taking is like on that platform.

At times, I learn a language by writing instructional materials. Having to explain something helps you learn it better. Of course, there are also times you learn something incorrectly and therefore share your inadequate understanding with others [33]. Nonetheless, I am considering trying to rewrite a few 151 or 207 lessons in Clojure and Python.

Somewhere in that, I find that playing with a language helps me learn it. Playing can me a host of different things. At times, I try to make things fail. At times, I try to understand how it behaves differently than languages I know. I recall experimenting with closures in Python when I was first learning Python and finding myself disappointed. I understand that things are better now [34].

In any case, I’ve written enough [35]. It’s time to go play with some languages [36].

---

[1] Why is it incorrectly named? Because it’s written with C#, and # is not the sharp sign.

[2] I see that I have a sketch of a musing about discovering Pharo. I may even finish that musing.

[3] I remember the joys of GHC, the Glasgow Haskell Compiler, also known as the Glorious Haskell Compiler. I see that it still exists.

[4] As an academic, I have a curriculum vitae rather than a résumé. I don’t list languages on my CV.

[5] I seem to recall something about Hilary Putnam’s claims about such an issue.

[6] Sandi Metz’ Practical Object-Oriented Design in Ruby also had an impact.

[7] In simple terms, lazy evaluation is an approach in which you only evaluate as much of something as you need. In this case, although we’re using an infinite list of integers, since we only need the first twenty-five, we only build the first twenty-five.

[8] Streams are like lists that you build on demand. In some languages, they provide a compromise between pure lazy evaluation and more standard evaluation models.

[9] I look forward to any comments my fifth-year colleagues [10] may have on Clojure.

[10] Or anyone else, for that matter. It’s just that my fifth-year colleagues are more likely to (a) read my musings and (b) have an opinion on the matter.

[11] The Java Virtual Machine.

[12] Youngest Son would probably say that it transpiles to JavaScript. I’m not sure how I feel about the rewriting of a long-standing poorly-chosen name [14].

[14] Computer scientists tend to use the term compile for translate a program from a high-level language to an executable, or something like that. It’s a horrid term, since compile traditionally means put things together. If I understand my computing history, compile was usually the last step of the translation; you joined together translated high-level language files with libraries to create the executable. And even that simplifies things a bit. In any case, it seems strange to use transpile instead of translate.

[15] And, it appears, transpiles to JavaScript.

[16] It appears he’s correct. A quick Web search reveals a seemingly-valid article that makes the claim and a Kotlin page on the official Android developer site. Ah, here’s the real deal: the announcement on the official Android Developers Blog [17].

Today [18] we’re announcing another big step: Android development will become increasingly Kotlin-first. Many new Jetpack APIs and features will be offered first in Kotlin. If you’re starting a new project, you should write it in Kotlin

[17] I would write that as Android Developers’ ’blog, but what do I know?

[18] 7 May 2019.

[19] Integrated Development Environments.

[20] Of course, if we switch from Java to something else in our Data Structures and Algorithms course, the question of Java IDE will no longer be important.

[21] Syntactic sugar is a term for language constructs that don’t enhance the capabilities of the language [22], but make it more convenient to write some things.

[22] Yes, dear readers, I do know that the Church-Turing thesis suggests that all languages are equally powerful. But there’s a difference between the addition of an object model and a more concise way to represent, say, setters and getters. I’d call the addition of objects an enhancement to the capabilities of the language and a concise way to add setters and getters a form of syntactic sugar.

[23] Of course, a lot of programmers really like Python and that doesn’t mean that it’s a beautiful language. I think there was even a time that a lot of programmers liked Perl [24].

[24] I still like Perl, at least for certain tasks.

[25] I realize that some of my colleagues may object to using commercial products. However, I see that there is a community version of IntelliJ and, well, it appears that the commercial/community pairing is becoming almost inevitable in many FLOSS endeavors [26].

[26] Racket appears to be primarily open. However, recent discussions on the Racket Users mailing list suggest that the way Racket was developed means that some licensing issues are unclear.

[27] Somewhere in my unorganized collection of notes for musings, I have notes on why I dislike the Ruby tutorial that we tend to use for our software design course.

[28] Unfortunately, I haven’t done enough Ruby programming since then, so lots of that knowledge has faded a bit.

[29] I realize that deepening my knowledge of Liquid may have damaged my brain. Liquid is an unpleasant language.

[30] Except when I read them on an electronic device.

[31] Students: Student membership in ACM is fairly cheap ($19/year) and, at last count, gave you access to Safari.

[32] I may have to take back my comments about the joy of having Safari as an ACM benefit. I’ve just spent fifteen minutes trying to get that benefit to work correctly, and the FAQ does not tell me what to do when I get a message like, Your membership was canceled, Samuelrebelsky. Please contact your administrator or O’Reilly Support.

[33] If I recall correctly, Experiments in Java, which I wrote about two decades ago, describes the import statement incorrectly. At the time, I was wedded enough to C that I thought that import was much like include. However, import is more about namespaces and linking than including code.

[34] More on that issue in a subsequent musing.

[35] Perhaps I’ve written too much.

[36] And environments.

---

Version 1.0 of 2019-09-09.