beachdog's Journal: Building Embedded Linux Systems a comment heavy mini-review

• Building Embedded Linux Systems, Second Edition

by Karim Yaghmour, Jon Masters, Gilad Ben-yossef & Philippe Gerum, published by O'Reilly, $49.99 c. 2008.

This is a mini-review with lots of comment.

I have a couple home projects in mind and I looked at Building Embedded Linux Systems to determine if this book is the guide I need.

Building Embedded Linux Systems is mostly about setting up a cross compiler environment and describing the various software changes available to make desktop Linux much smaller, boot faster, run in real time and run with various kinds of volatile and non-volatile memory and do various kinds of networking and communication.

The strength of Building Embedded Systems is if you do any kind of software development for any kind of Linux embedded system, and if you are employed at this work, this book will say enough things to be worth your while.

Building Embedded Systems is not a book about system design and it is also not a book about specific embedded system hardware. The book has the editorial focus and intensity of other O'Reilly books... this is a book mostly about putting together the development environment, file system choices and compiler outputs that are loaded onto a hardware board to make a Linux based embedded system.

These are the weaknesses, reasons why on my modest projects I am unlikely to add Building Embedded Systems to my library:

* For my spinning clothes dryer I will continue to use a Basic Stamp with some thermistors, power transistors and push button switches. Embedded Linux still isn't needed for a project with one measured value, one motor and no more than $.30 per day energy savings.

* For my autonomous electric wheelchair conversion project, I am going to stick with my Dell Inspiron 1200 running Ubuntu. This is a project where finding and integrating software pieces is the real problem. I already have a working web cam and a working GPS and a working wireless link and a working USB-serial port.

For the autonomous electric vehicle project I see doing embedded Linux as trying to do hardware development and software development at the same time. Embedded development puts me in the situation of dealing with hardware problems and operating system software and application software development at the same time.

So my feeling about Building Embedded Systems is not so much a criticism of the book as a comment on the state of the art. Presently embedded systems require a rather elaborate setup: You have to commit to a specific target hardware. Then you have to assemble a dedicated development system to write and compile software builds for the target system. Then you have to be prepared to make further investments in hardware and software debugging.

So for me, I see the better path for my projects is to avoid formal embedded computing.

I will use the Basic Stamp I have on hand for simple switch and motor control and feed back loops. If I need more micro controllers, I will get some Arduino devices, they have a simple A-D input and open source software and some boards support a USB interface.

For the autonomous wheelchair project, I'll develop using a laptop. I feel once I have a running and testable software stack, the design problem will be to study adding solar panels and reducing energy usage. Migrating to a tiny or embedded CPU will be a less important design priority.