Книга: Distributed operating systems

9.6. COOL: AN OBJECT-ORIENTED SUBSYSTEM

The UNIX subsystem is really nothing more than a collection of Chorus processes that are marked as a subsystem. Consequently, it is possible to have other subsystems running at the same time. A second subsystem that has been developed for Chorus is COOL (Chorus Object-Oriented Layer). It was designed for research on object-oriented programming and to bridge the gap between coarse-grained system objects, such as files, and fine-grained language objects, such as structures (records). We will describe COOL in this section. For more information, see (Lea et al., 1991, 1993).

Work on the first version, COOL-1, began in 1988. The goal was to provide system-level support for fine-grained object-oriented languages and applications, and do so in such a way that new COOL programs and old UNIX programs could run side by side on the same machine without interfering.

Each Object in COOL-1 consisted of two Chorus segments, one for the data and one for the code. Programs did not access the data segments directly. Instead they invoked procedures, called methods, located in the objects' code segments. In this way, the objects' internal representation was hidden from the user, allowing object writers the freedom to use whatever representation seemed best (for example, an array or a linked list). This representation could even be changed later, without programs using the objects even knowing. Multiple objects of the same class shared the same code segment, to save memory.

To make a long story short, the resulting system was disappointing in terms of performance, resource usage, and so on, and the gap between coarse-grained system objects and fine-grained language objects was not bridged. In 1990, the designers started over and designed COOL-2. This system was running a year later. Below we will describe its architecture and implementation.

Оглавление книги