Logical shift Project home

Project specification

A lightweight, connected language for small devices

The problem

To design a language suitable for execution on a wide range of consumer mobile devices.

Elaboration

'Consumer mobile devices' includes mobile phones and PDAs. They are assumed to have some form of outside data connectivity. These devices have several traits that have an important impact on the design of the language:

• User interface:
  • Small screen, usually black and white graphics only, with limited text display ability, but greyscale and colour devices possible.
  • Very variable input facilities. Minimum is telephone keypad, with menu navigation buttons, but may extend to full keyboard or even a touch screen with handwriting recognition.
  • 'Push' operations, such as SMS or receiving phone calls, are often a more compelling reason for owning a mobile communications device, than their 'pull' equivalent.
• Technical:
  • Very little processing power in most of these devices, although low power/high performance chips such as ARM offer the prospect of much better performance in the future.
  • 'Interesting' memory architechtures further limit program design. While overall memory may actually be large, dynamically allocatable RAM may be limited and call stack size may be limited further still.
  • Power consumption is a big issue in many of these devices. Often they power down into a sleep mode when they are not doing anything, and may cease to provide non-essential events (for example, timeouts) to software.
  • While these devices may offer the ability to switch between many applications, they may not provide any multitasking at the operating system level, leaving that as an illusion to be maintained by the individual applications.

The most important question associated with any such application as this is: "What will it be used for?". In my opinion, a mobile language will be most useful for content providers to add interactivity to their content. Aside from the obvious mobile games applications, a stock company, for instance, that might presently send important alerts as SMS messages or emails, may use a program running on the users mobile device to display graphs and other statistics, and offer the user the option to request more information, or perhaps buy or sell shares.

Objectives

These objectives are specified in roughly chronological order. Objectives considered optional are also specified, and may be qualified with 'preferred' if their completion is to be preferred for the completion of the project.

• Provide a definition of the range of devices that the language will run on
• Define the range of services the language will provide
• Define the language semantics
  • Define a low-level bytecode for the language
  • Define a high-level language for writing applications
• Define a standard library of routines for the language
• (Optional, Preferred) Write a bytecode assembler
  • Define assembly language
  • Write a BNF grammar for the bytecode
  • Create a parser and code generator
• (Optional, Preferred) Implement a bytecode interpreter
  • Define assumed device limitations
  • Define hardware/OS abstraction layer
  • Create interpreter and supporting software
• (Optional, Preferred) Write a language compiler
• (Optional) Design an interpreter intended to allow the design of short programs on mobile devices

Methods

For this project, C is the language of choice. It is implemented for a very wide range of devices and targets, and there are many retargetable compilers (such as lcc or gcc) that make exapanding the range easy. For the purposes of the project, I will use the Palm Pilot, OS 3.1, as the intended target, as development tools and emulators are freely available. Flex and Yacc are traditional compiler building tools, and may make a good choice for developing assemblers and compilers.

Timetable

(To be decided)

Resources

• Standard UNIX and GNU utilities - gcc, flex, yacc, perl etc
• GNU PalmOS development environment
• PalmOS emulator (I am a registered developer)

The intention is to do most of the development on my home machine, and maintain offsite backups of the data on the University system, as well as through my normal CD backup system. The University system will also be useful for providing demonstrations of working code.