Previous Contents Next

Presentation of part II

We describe the set of elements of the environment included in the language distribution. There one finds different compilers, numerous libraries, program analysis tools, lexical and syntactic analysis tools, and an interface with the C language.

Objective CAML is a compiled language offering two types of code generation:
  1. bytecode to be executed by a virtual machine;
  2. native code to be executed directly by a microprocessor.
The Objective CAML toplevel uses bytecode to execute the phrases submitted to it. It constitutes the primary development aid, offering the possibility of rapid typing, compilation and testing of function definitions. Moreover, it offers a trace mechanism visualizing parameter values and return values of functions.

The other usual development tools are supplied by the distribution as well: file dependency computation, debugging and profiling. The debugger allows one to execute programs step-by-step, use breakpoints and inspect values. The profiling tool gives measurements of the number of calls or the amount of time spent in a particular function or a particular part of the code. These two tools are only available for Unix platforms.

The richness of a language derives from its core but also from the libraries, sets of reusable programs, which come with it. Objective CAML is no exception to the rule. We have already portrayed to a large extent the graphical library that comes with the distribution. There are many others which we will describe. Libraries bring new functionality to the language, but they are not without drawbacks. In particular, they can present some difficulty vis-a-vis the type discipline.

However rich a language's set of libraries may be, it will always be necessary that it be able to communicate with another language. The Objective CAML distribution includes an interface with the C language allowing Objective CAML to call C functions or be called by them. The difficulty of understanding and implementing this interface lies in the fact that the memory models of Objective CAML and C are different. The essential reason for this difference is that an Objective CAML program includes a garbage collection mechanism.

C as well as Objective CAML allow dynamic memory allocation, and thus fine control over space according to the needs of a program. This only makes sense if unused space can be reclaimed for other use during the course of execution. Garbage collection frees the programmer from responsibility for managing deallocation, a frequent source of execution errors. This feature constitutes one of the safety elements of the Objective CAML language.

However, this mechanism has an impact on the representation of data. Also, knowledge of the guiding principles of memory management is indispensable in order to use communication between the Objective CAML world and the C world correctly.

Chapter 7 presents the basic elements of the Objective CAML system: virtual machine, compilers, and execution library. It describes the language's different compilation modes and compares their portability and efficiency.

Chapter 8 gives a bird's-eye view of the set of predefined types, functions, and exceptions that come with the system distribution. It does not do away with the need to read the reference manual ([LRVD99]) which describes these libraries very well. On the contrary it focuses on the new functionalities supplied by some of them. In particular we may mention output formatting, persistence of values and interfacing with the operating system.

Chapter 9 presents different garbage collection methods in order to then describe the mechanism used by Objective CAML.

Chapter 10 presents debugging tools for Objective CAML programs. Although still somewhat frustrating in some respects, these tools quite often allow one to understand why a program does not work.

Chapter 11 describes the language's different approaches to lexical and syntactic analysis problems: a regular expression library, the ocamlex and ocamlyacc tools, but also the use of streams.

Chapter 12 describes the interface with the C language. It is no longer possible for a language to be completely isolated from other languages. This interface lets an Objective CAML program call a C function, while passing it values from the Objective CAML world, and vice-versa. The main difficulty with this interface stems from the memory model. For this reason it is recommended that you read the 9 chapter beforehand.

Chapter 13 covers two applications: an improved graphics library based on a hierarchical model of graphical components inspired by the JAVA AWT2; and a classic program to find least-cost paths in a graph using our new graphical interface as well as a cache memory mechanism.

Previous Contents Next