We are all natives now, and everybody else not immediately one of us is an exotic. What looked once to be a matter of finding out whether savages could distinguish fact from fancy now looks to be a matter of finding out how others, across the sea or down the corridor, organise their significative world. (Clifford Geertz)
The Java Programming Language turned 25 years old on May 23 2020.
I’ve been writing Java code for 24 years, since 1996.
When Java arrived on the scene, my languages of choice were C, C++ and Perl, but mostly the first and last.
For me, there were at first various BASIC dialects, with smatterings of assembly code. Then at university and just at the moment I thought I knew something about programming, I learned Pascal, C and Fortran.
Later, languages such as LISP, Prolog, ML, CLIPS, SQL, even COBOL (about which I really never could find anything to like) showed me different ways in which a computer could be programmed.
Programming paradigms and language translation (compilers, interpreters) have interested me since around 1989. I wrote the ACE BASIC compiler in C from 1991 to 1996.
I’ve since designed and implemented other programming languages, including LittleLisp for the Newton PDA, and others that were only ever experiments, most personal projects, one commercial. In recent times I’ve been developing a domain specific functional language called VeLa that I’ll write about some time.
From the start, Java made it possible to write cross-platform programs, with or without GUIs, with relative ease.
Its byte-code compilation model resonated with me after having written my first compiler for a subset of Pascal in 1989 that generated USCD p-codes.
In 2004, six years after Java was released, Java 5 brought generics, regular expressions and assertions. I had craved these, especially the first two, for years, since writing C++ and Perl code in the early to mid-90s.
Java’s history is something of a long strange trip: from set-top boxes to web applets, the desktop, enterprise frameworks, smart cards and other embedded devices, to Just In Time compilation and dynamic profiling, WebStart and security model criticisms, and from Sun Microsystems to Oracle.
C# learned from Java’s mistakes. Some of its proponents have not always seemed to me to be honest about acknowledging its heritage though.
Programming Languages are tools for thought, for human communication, and not merely a means by which to bend a machine to our will. As I’ve mentioned elsewhere, they can also determine what it is possible for us to think.
My working days are now dominated by R, Python, C++ and sometimes Fortran; and bash of course. R and Python are powerful and have a rich ecosystem of tools and libraries, but I’m no fan of dynamic typing which makes it easy to get started, yet leaves too much unsaid, too many possibilities for error. This problem only becomes worse as programs grow.
Rust, Julia, Go, Clojure, Swift are all contenders to the throne now too. Which of these will stand the test of time? Which will one day be considered to have legacy code bases, just like Java, C, C++, Fortran, COBOL? All of them, I would say.
Then there are those languages like Haskell that blaze trails but tend not to be dominant. Java and most of the current crop owe such languages a debt of gratitude as functional programming’s higher order functions, type inference, and immutability have come out of academia to take over the world.
Distinguishing fact from fancy, fashion from solid ground, making predictions about the future of the programming language landscape is no easy task.
The older I get, the more selective I become about what new languages and frameworks to learn and the more interested I am in as much static typing, type inference, and run-time efficiency as I can get my hands on.
Yet, after a quarter century, I still like to code in Java whenever I can. It has its own beauty. I don’t pay much attention to its detractors. Of course, that’s where it’s important to think about context. If I’m writing code for a tiny embedded system or want to squeeze the most out of a machine, I’m more likely to code in C or C++ than anything else. But if I want to build a big habitable castle in the sky, I’ll still choose Java when I can.