A look at two new languages: Vala and Clojure

Some programming languages forge new ground, others are refinements of previously existing ideas, and still others tackle a specific problem in a new and better way. This article will look at two up-and-coming languages that are not yet as widely adopted as C, C++, or Java, but offer developers some intriguing benefits: Vala and Clojure. Vala is designed specifically to build native applications for the GNOME desktop environment, and Clojure brings the power of Lisp and functional programming to the Java Virtual Machine (JVM).

Vala

GNOME itself is written in C, but many of the higher-level libraries that make up the GNOME platform (such as GTK+ and Pango) make heavy use of the GLib and GObject frameworks, which implement a complete object system including types, messaging, classes, and all of the other components that make up a typical object oriented application program interface (API). GObject makes it possible to write object oriented GNOME applications in C, and also makes it simple to provide wrappers to GNOME for object oriented languages like C++ or Python.

While GObject makes object-oriented programming in C possible, it remains somewhat cumbersome. The GObject wrappers for other languages can be even worse, introducing yet another layer of dependencies which can include virtual machines of their own. Vala was created in an attempt to address these problems and make GNOME programming easier. Vala provides a C#-like syntax that integrates GObject features, but the valac compiler compiles to C output rather than assembly. The resulting C compiles normally with gcc, plus it can be distributed as source packages fully usable on platforms that do not have Vala installed. That allows the developer to use Vala's modern language features — dynamic type system, assisted memory management, and so on — while still making source releases compatible with C applications and libraries.

In addition to GLib and GObject, the Vala project produces bindings to the major GNOME libraries: GTK+, GDK, Cairo, SDL, GStreamer, D-Bus, GIO, and many more. In addition, several external projects provide their own bindings for Vala (such as Clutter), and there are third-party efforts to create bindings for other libraries, such as OpenGL. Plugins for Vala support are available for the Eclipse, Anjuta, and MonoDevelop IDEs, as well as for the editors Emacs, Vim, and GEdit. Val(a)IDE, still in development, is a dedicated Vala IDE written in Vala itself.

The latest release of Vala is 0.7.3, which went public on May 26, 2009. It is a bugfix release for the 0.7 series, which is intended to be the last development series before 1.0. The biggest change in the 0.7 series, according to developer Jürg Billeter, is to C header file generation. In previous releases, Vala would generate a .h header file for every .vala source file, which caused problems:

That change demanded making changes to the build system for existing Vala applications, but Billeter said it should be the last such headache for a while.

Vala 1.0 is scheduled to be released in September of 2009, coinciding with the release of GNOME 2.28. According to Billeter, the project is on track to make the 1.0 deadline, which will mark the start of a long stable release cycle during which application developers can count on language stability. The bindings to GNOME libraries are maintained separately, Billeter said, and are added to and enhanced on an as-needed basis, so they are not subject to the same schedule.

The project offers a tutorial and sample code for those new to Vala; some familiarity with GObject is expected. Benchmarks between Vala and C show no significant execution speed losses (in some cases, Vala even beats plain C).

The project maintains a list of applications developed using Vala, in whole or in part, including GNOME panel applets and full-fledged applications. The webcam utility Cheese is currently maintaining a Vala branch in addition to its main trunk, just to road test the viability of the language. Billeter noted that although people generally think of using Vala only when starting a new project, it is possible to incorporate it into an existing C code base, too. "Among our users are both, people writing C that now want to move to something higher level and also people that just start programming in the GNOME environment," he said.

Clojure

While Vala is a language designed around a particular object system, targeting a particular desktop environment, Clojure could not be more different. It is a dialect of the functional-flavored programming language Lisp, implemented on the Java platform. That makes it cross-platform; Clojure applications are compiled to Java bytecode, so they can run on any platform with a well-supported JVM.

Creator Rich Hickey has explained that building on top of the JVM grants Clojure automatic platform stability from a broad user and developer community, but that itself was not the goal of creating the language. Hickey's primary interest was concurrency — he wanted the ability to write multi-threaded applications, but increasingly found the mutable, stateful paradigm of object oriented programming to be part of the problem. "Discovering Common Lisp after over a decade of C++, I said to myself — 'What have I been doing with my life?', and resolved to at some point code in Lisp. Several years of trying to make that practical, with 'bridges' like jFli, Foil etc, made it clear that bridges weren't going to be sufficient for the commercial work I was doing."

Hickey became less enamored of object oriented programming and started adopting a functional-programming style in his work, which he found to make the code more robust and easier for him and for his coworkers to understand. Eventually, maintaining that style in other languages like C# became more trouble than it was worth:

Clojure does provide persistent data structures, although it does considerably more. For those unfamiliar, functional programming (the style from which Lisp and Clojure originate) places a greater emphasis on functions as first-class objects, meaning that functions can be placed into data structures, passed as arguments to other functions, evaluated in comparisons, even returned as the return value of another function. Moreover, functions do not have "side effects" — the ability to modify program state or data. This paradigm focuses on computation in the mathematical sense, rather than procedural algorithms, and is a completely different approach to programming.

As a language, Clojure is a Lisp-1, part of the same family of Lisp variants as Scheme, notable for sharing a single namespace between functions and variables. Clojure differs from Scheme and other Lisp dialects in several respects documented at the Clojure web site. For application developers, the most significant distinction is that Clojure defaults to making all data structures immutable. To maintain program state, developers must use one of four special mutable structures that are explicitly designed to be shared between threads: refs, vars, atoms, and agents. Clojure uses software transactional memory (STM) to coordinate changing these mutable structures while keeping them in a consistent state, much like a transactional database. This model makes it considerably simpler to write thread-safe code than it is in object oriented languages. No locks are required, therefore there are no deadlocks or race conditions.

Like other Lisp implementations, Clojure is interpreted through a console-like read-eval-print-loop (REPL). The user launches the REPL from a .jar file and is presented with the REPL command prompt, from which he or she can load Clojure programs or directly write and execute functions and macros. The code is compiled on-the-fly to Java bytecode, which is then in turn executed by the JVM. The REPL environment is much like an interactive IDE and debugger all rolled into one, but for distribution purposes, Clojure code can be compiled ahead of time into ordinary Java applications. Because it is hosted by the JVM, Clojure can automatically make use of its features, including the type system, thread implementation, and garbage collector, rather than having to re-implement each of them. Clojure code can also call Java libraries, opening up a wealth of classes and interfaces to Clojure programmers.

Clojure 1.0 was released on May 4, 2009. There are several good resources online for learning about the language and for getting started, although a general introduction to Lisp is probably warranted for those with no experience in functional programming. Though there are not large-scale projects using Clojure, an active community is growing around it, including several local users' groups. The Clojure site offers documentation of the language syntax and examples (including example code), there is a very active Google Groups discussion forum, and Mark Volkmann's Clojure page at Object Computing tracks articles, slides, and wikibooks about Clojure.

Two-way interoperability

Vala and Clojure seem to have little if anything in common; one is object oriented and the other functional, one aimed at a specific desktop system and the other intentionally cross-platform. They are kindred spirits in one sense, however — they seek to build a more modern, robust language implementation on top of an existing, established platform. Vala's goal is to let C programmers more easily take advantage of the power of GObject and GNOME, and Clojure's is to let developers easily write concurrent applications on top of the stability of the JVM.

What is equally important is that both projects maintain bi-directional compatibility with their underlying languages and platforms. A Vala program can use any C library, and a C program can use any library written in Vala. Likewise, Clojure code can be compiled to Java, and Clojure applications can use any Java class or interface. Such interoperability will likely increase adoption of both of these languages, and it is a welcome sight in any project.