Overview

Modern desktop development environments use remarkably complex translation techniques. Source code is seldom translated directly into loadable binary images. Sophisticated suites of tools translate the source into relocatable modules, sometimes with and sometimes without debug and symbolic information. Complex, highly optimized linkers and loaders dynamically combine these modules and map them to specific memory locations when the application is executed.

It s amazing that the process can seem so simple. Despite all this behind- the-scenes complexity, desktop application developers just select whether they want a free-standing executable or a DLL (Dynamic Link Library) and then click Compile. Desktop application developers seldom need to give their development tools any information about the hardware. Because the translation tools always generate code for the same, highly standardized hardware environment, the tools can be preconfigured with all they need to know about the hardware.

Embedded systems developers don t enjoy this luxury. An embedded system runs on unique hardware, hardware that probably didn t exist when the development tools were created. Despite processor advances, the eventual machine language is never machine independent. Thus, as part of the development effort, the embedded systems developer must direct the tools concerning how to translate the source for the specific hardware. This means embedded systems developers must know much more about their development tools and how they work than do their application-oriented counterparts.

Assumptions about the hardware are only part of what makes the application development environment easier to master. The application developer also can safely assume a consistent run-time...