Embedded Muse 167 Copyright 2008 TGG October 23, 2008

You may redistribute this newsletter for noncommercial purposes. For commercial use contact jack@ganssle.com. Subscribe and unsubscribe info is at the end of this email.

EDITOR: Jack Ganssle, jack@ganssle.com

- Editorís Notes
- Book Review
- The Failure of Reuse
- Jobs!
- Joke for the Week
- About The Embedded Muse

This issue of The Embedded Muse is sponsored by Netrino.

Are you an embedded C expert? Know the difference between volatile * const and const * volatile? See how your skills measure up in a fun, quick, online quiz. Challenge your peers for bragging rights. All takers are automatically registered to win one of three iPod Nanos.

Start at http://www.netrino.com/Embedded-Systems/Embedded-C-Quiz-Muse

Editorís Notes

Did you know it IS possible to create accurate schedules? Or that most projects consume 50% of the development time in debug and test, and that itís not hard to slash that number drastically? Or that we know how to manage the quantitative relationship between complexity and bugs? Learn this and far more at my Better Firmware Faster class, presented at your facility. See https://www.ganssle.com/classes.htm .

case San Jose:
case Phoenix:
case Austin:

Are you in the San Jose, Phoenix or Austin areas? Iíll present a public version of the Better Firmware Faster class in San Jose on December 8, Phoenix on December 10 and Austin on December 12. Registration and other info here: https://www.ganssle.com/classes.htm . Youíll earn 0.7 Continuing Education Units, learn a lot, and have more than a little fun. Sign up before November 8 and receive a $50.00 discount.

Harold Teague has an interesting question for Muse readers: Weíre in the process of establishing job roles along the lines of firmware architect, firmware design engineer, and firmware integration/systems engineer. Does anyone have these or similar descriptions already defined?

Let me know what you think and Iíll publish the results here.

In the seemingly never-ending quest for the perfect communications monitor, Bill Eubank sent info about an RS-232 monitor that runs on a Game Boy! See http://www.databoy.netfirms.com/. He writes: I have had one of these for may years and I still use it regularly. It monitors both sides of the communications, receive and transmit. It can also capture this information and transfer it to a PC for storage and/or printing. I met Yosuf Taraki, the designer, several years ago. He's a very nice guy.

Book Review

After the last Great Book Giveaway, J. Silva submitted a review of "Practical Software Estimation":

Function point estimation methods are usually glossed over in a small section in program management books. ďPractical Software EstimationĒ by M. A. Parthasarathy provides an in-depth discussion of function point-based methodologies for program management.

Unfortunately, with all these helpful examples, it was difficult to synthesize much of the information into something useful in real life for a number of reasons. First, the process seems so complicated that it is difficult to apply it to some projects. From the high level, the process is simple but the devil is in the details. Second, it seems time consuming. I do not think this method would work for small, fast projects.

Third, it requires the project manager to not only have a strong understanding of the requirements, but also how the design would breakdown. It is difficult to discern if this is a fault with the actual method or the authorís presentation.

I would recommend this book to a program manager on a large-scale, complex government contract with a long time horizon and a need for accurate schedule and costs. I would not recommend this for startup organizations, smaller project developers, or novice program managers.

The Failure of Reuse

Reuse is the holy grail of software engineering, one that is so entrenched in our belief system no one dares to question its virtue. The quest for reusable components is one of the foundations of object oriented programming and all of the tools and languages that OOP has spawned.

Yet, my observations suggest that at least in the embedded space reuse has been a dismal failure.

First, let me define ďreuse.Ē We can talk about carrying-over code, or salvaged code, both of which imply grabbing big source files and beating them into submission till a new product appears. Not to knock that; itís much better than starting from scratch each time (but see my comments towards the end). We reach the nirvana of reuse, though, when we leave the source code alone, when weíre able to pluck a module from the virtual shelf and drop it in to a new project, unchanged.

Martin Griss and others have observed that a module isnít really reusable till it has been reused three times. No matter how good our intentions, the first time we try to reuse something we discover a facet of the new problem the old module just canít manage. So we tune it. This happens a couple of times till the thing is generally reusable. Thatís not because weíre stupid; itís simply because domain analysis is hard. No one is smart enough to understand how a function might get used in other apps.

Itís expensive to do a forward-looking design of a function or module. Youíll always save money Ėin the short term Ė solving todayís very specific problem while ignoring the anticipated demands of future projects. If you elect to pursue a careful program of reuse your projects will initially come in late and over-budget.

Reuse is hard. Itís like a savings account. My kids complain that if they stick a few bucks in the bank then thatís money they canít useÖ and itís just a piddling sum anyway. Sure. The value of savings comes after making regular deposits. Ditto for reuse. The cost is all up-front, the benefits come from withdrawals made in later years.

Sure, we all know the long-term outweighs todayís concerns. (Though the recent financial meltdown suggests thatís a lesson some CEOs have missed). Most bosses buy into the idea of the benefits of reuse, without being willing to stand the pain of creating the reusable components. And thatís the rub. When the ship date looms closer, too many bosses will tell us to toss out any sort of discipline that has long-term benefits in pursuit of a near-term release. So, in practice, reuse often fails since schedules generally dominate over any other parameter.

I also suspect most of us donít want to reuse code. 35% of RTOSes are homebrew, despite at least 100 commercial Ė free and otherwise Ė products on the market. Nothing is easier to beg, borrow or buy than an RTOS, yet most of us still refuse to practice even this most simple of all reuse strategies. Why is this? Iíve heard many specious arguments (too expensive, yet there are plenty of freebies; we donít trust the code, but some are safety certified, etc), plus a few really good arguments (itís all legacy code, no one is willing to make risky changes).

I think aggressive reuse is our only hope of salvation from the morass of expensive and unreliable code. Building correct firmware is so difficult that we and our bosses have a fiduciary responsibility to find ways to make it reusable.

But weíre not doing it. Sure itís hard. Yes, itís initially expensive. And of course we cannot reuse everything; a lot of what we build will always be inherently unreusable, like hardware drivers that get tossed with each new spin of the design.

So whatís the deal? Why is reuse such a failure? Is the best we can do, or chose to do, is to salvage old source files?

A closing comment: In ďEmpirically analyzing software reuse in a production environment,Ē Selby et al showed that hacking source has quickly-diminishing benefits. Once one changes more than about a quarter of the code thereís not much cost saving over starting from scratch.


Joke for the Week

Guillermo Zepeda-Lopez sent this:
There is an effort in our company to find ways to make our automotive ECUs lighter. Two of my SW colleagues (Ricardo Garcia and Antonio Lozano) turned in some great ideas to reduce weight through the SW in our product:

- Write "lighter" code.
- Declare all variables as "volatile".
- Recursive calls to function ReduceWeight(int CurrentWeight).
- Use more powerful microcontrollers so that we reduce the CPU "load".
- Delete all comments from the code.
- Use an operating system with a bigger "footprint". That way, we distribute the weight on a bigger area (less pounds per inch).
- Leave all I/O pins "floating".
- Always use "Tiny" memory model.
- Stop using processes. Use only "light-weight processes".
- Do more "on-the-fly" processing.
- Stop drawing state diagrams with rectangles, use "bubbles" instead.
- Don't eliminate all bugs. Remember some bugs can fly.