Dr. Dobb's, in cooperation with Forrester Research, wants to better understand how evolving tools and technologies are affecting the software developer community. To help gain this understanding, we would appreciate it if you would take a few minutes to complete the Forrester-Dr. Dobb's 2009 Developer Technographics Survey. It should take no more than 5-10 minutes to complete the survey.

The first 100 readers to complete to the survey will receive a Dr. Dobb's DVD Release 5 -- a $99.95 value. Furthermore, all respondents who complete the survey will be entered into a drawing for one of five $50 gift cards. In addition, all respondents who complete the survey will also get a copy of the Forrester report, which summarizes the results -- a $750 value.

Your responses will only be reported in aggregate form. Although you do not need to provide your name and email address, you will have to provide this information in order to receive the complimentary Forrester report, Dr. Dobb's DVD or be registered for the gift card drawing.

The deadline for completion is July 17th, 2009. We plan on publishing the results in July/August 2009. Thanks in advance for participating in this unique and important survey.

Please let me know if you have any questions, and thank you in advance for your time.

-- Jonathan Erickson
jerickson@ddj.com

If your idea of a good time is 72 hours of non-stop programming, then with any luck, you were able to participate in the ICFP 2009 Programming Contest, part of the upcoming International Conference on Functional Programming.

Hosted this year by the University of Kansas and sponsored by the ACM, the contest had more than 800 teams around the world moving simulated satellites around in space. According to contest chair and KU professor Andy Gill, more than 300 teams solved at least one problem this year. Even though the contest was associated with a functional programming language conference, there was no requirement that entries had to be with functional languages. Last year's winner was in Java, with Ocaml taking the honors in a lightning round. Other languages that have done well in the past include everything from Haskell and Python to C and C++.

This year, the contest consisted of four problems, each requiring contestants to write programs that controlled a satellite. For each problem, there was a series of scenarios, each scenario representing an instance of the problem with the satellites in a different starting configuration. In every case, contestants controlled the satellite by providing input to a thruster actuator.

• The Hohmann problem required the controller to transfer the satellite, initially in a circular orbit around Earth, into a different circular orbit. The satellite had to remain within 1km of the target orbit radius for 900 seconds.
• In the Meet and Greet problem, contestants moved a satellite in a circular orbit about Earth to meet with a second satellite, also in a circular orbit. The target satellite had to maintain its orbit throughout the simulation. The position of the controlled satellite had to remain within 1km of the target for 900 consecutive seconds.
• In the Eccentric Meet and Greet, contestants moved the satellite from one orbit to meet with a target satellite. However, the orbits of both satellites were in arbitrary ellipses. Again, the controlled satellite had to remain within 1km of the target for 900 consecutive seconds.
• The Operation Clear Skies problem required contestants to control the satellite so that it visited a collection of 11 target satellites.

For a complete description of the problem specification, go here.

So who out maneuvered whom to win the contest? We don't know yet. The winners will be announced at the ICFP Conference in Edinburgh, Scotland, at the end of August. By then, you'd hope, all of the contestants will have caught up on their rest.

-- Jonathan Erickson
jerickson@ddj.com

Does anyone still use dot-matrix printers? I don't, and I can't recall the last time I saw one. Wait. I recently bought a replacement brake-light bulb and my purchase receipt was printed out on a dot-matrix printer. And now that I think about it, my bank prints out deposit and withdraw receipts on small dot-matrix printers.

In fact, say researchers at Saarland University, 30% of the banks surveyed still use dot-matrix printers. Of that number, 29.9% use dot-matrix for receipts like the one's I get, and 83% use them for other purposes. Anyone else? How about doctor's offices? What the same researchers found was that 58.4% of the clinics polled still used dot-matrix printers, 79.4% for prescriptions and 84.5% for other uses.

But do we really care whether these businesses choose to use technology from an earlier era? We should. Why? Because it is a security issue, at least according to Saarland University's Michael Backes, Markus Durmuth, Sebastian Gerling, Manfred Pinkal, and Caroline Sporleder. What the researchers discovered is that by capturing (recording) the ratta-tat-tat of dot-matrix printers, then applying feature extraction from speech-recognition (Hidden Markov Models) and music processing, you can extract valuable, private data from dot-matrix printers. Little things like your account number, account balance, health issues, and the like. Stuff you thought was private. With permission of course, the researchers successfully mounted an attack in a doctor's office and recovered the content of a medical prescription. According to their paper How Printers Can Breach Our Privacy: Acoustic Side-Channel Attacks On Printers, the attack was conducted under realistic conditions with people talking in the waiting room.

So why do people still use dot-matrix printers? For all the reasons you'd expect -- they're cheap, durable, and still work with old hardware and computer systems. Consequently, only 4.7% of the doctor's offices and 8.3% of the banks surveyed had plans to upgrade printers.

My bank account (such as it is) is one thing. But having my brake-light replacements crop up on the Internet is something else altogether. I'm off to alert the auto-parts store.

-- Jonathan Erickson
jerickson@ddj.com

Computer vision is all about machines that can see. Artificial systems that get information from images. And it is the subject of Richard Szeliski's upcoming book entitled Computer Vision: Algorithms and Applications.

Actually, it's not quite a book yet. It's an almost-a-book (beta book?) that's based on a series of lectures Szeliski, who is a researcher at Microsoft Research, presented at the University of Washington. Szeliski begins with a description of basic concepts on image formation and processing. He then covers advanced topics in feature detection and matching, segmentation, calibration, structure from motion, image stitching, computational photography, stereo, recognition, and image-based rendering.

Szeliski is looking for feedback, and to get it, he's made a draft of the book freely available here. Download and enjoy -- and share your comments by sending him feedback.

In case you wondered, Szeliski isn't a lab coat type of researcher. A lot of his research finds its way into real products. For instance, Szeliski was one of the main people behind Photosynth, software that lets you create panoramic images from collections of digital photos, Digital Image Pro, and Windows Live Photo Gallery (Beta) software.

To top it off, Szeliski is a founding editor of Foundations and Trends in Computer Graphics and Vision. Ah...a man after my own heart.

-- Jonathan Erickson
jerickson@ddj.com

I have a gift that I cherish. Actually, two gifts now that I think of it. Gift #1: I can sleep on airplanes. When the wheels go up, I go out -- like a light. Gift #2: I don't suffer from jet lag. Take off from O'Hare and touch down in Frankfurt, and I'm rested and raring to go. My internal time clock and local environmental cues are in sync.

But not everyone is as blessed. For those of you who aren't so lucky, researchers at the University of Michigan have come to the rescue. They've written an application that implements a timed light-exposure schedule to avoid jet lag. The program, called "Shifter," considers inputs like background light level and the number of time zones you've traveled to re-sync your body with its new environment. Based on a mathematical model, Shifter gives you a schedule for when you should apply countermeasures (such as bright light) to reduce the effects of jet lag.

As described in Taking the Lag out of Jet Lag through Model-Based Schedule Design, Shifter is implemented in MATLAB 7.7. The schedule building blocks are implemented as MATLAB functions. Each building block is designed to be called with a variable number of parameters. The optimization interface is designed to use MATLAB's fminsearch function to build schedules with a variable number of parameters to be optimized.

Using their computational method, researchers Dennis Dean, Daniel Forger, and Elizabeth Klerman1 simulated shifting sleep-wake schedules and the subsequent light interventions for realigning internal clocks with local time. They found that the mathematical computation resulted in quicker design of schedules and also predictions of substantial performance improvements. They are able to show that the computation provided the optimal result for timing light exposure to reduce jet lag symptoms.

Yawn...I get tired just thinking about it.

-- Jonathan Erickson
jerickson@ddj.com

When California Governator Arnold Schwarzenegger announced plans to save money by replacing school textbooks with digital ebooks last month, there were enough guffaws to go around. But make no beans about it -- it will happen. Maybe not this fall for public schools in California, but eventually.

Not to be deterred, Schwarzenegger hopes to start with science and math digital ebooks for high-school students beginning this fall. We'll see. Content developers (aka, "textbook publishers" in a previous life) had until this week to submit their wares for evaluation. Once the ebooks are in place, websites will be put in place to accompany and update the ebooks. That's the plan anyway.

But California isn't alone in promoting ebooks in the classroom. Starting this fall, every first-year grad student in the University of Washington's Computer Science and Engineering Department will receive an Amazon Kindle DX ebook to use in place of printed textbooks and research papers. Moreover, approximately 40 students will receive Kindle-based ebooks and other required reading materials free of charge, in a pilot program under the direction of Professor Ed Lazowska.

Among the graduate-level classes that will using the ebooks are: Software Engineering, Concepts of Programming Languages, Design and Analysis of Algorithms, Computational Complexity, Principles of Database Systems, and Computer Systems Architecture.

The University of Washington isn't the only institution participating in the Kindle DX pilot program. Other schools include Princeton University, Case Western Reserve University, Reed College, Arizona State University, Pace University, and the University of Virginia. Researchers monitoring the pilot program hope to find out more about the strengths and weaknesses of ebooks relative to traditional content delivery.

If nothing else, students participating in the program should be able to stand a little taller, what with one ebook instead of half-dozen or so textbooks weighing down their backpacks.

-- Jonathan Erickson
jerickson@ddj.com

Business can make for some strange bedfellows. For instance, it hasn't been that long since Microsoft and Novell were going at it tooth-and-nail, as Andrew Schulman documented in the classic 1993 article Examining the Windows AARD Detection Code where he unraveled Microsoft's attempts to keep betas of Windows from running on Novell's DR DOS. 1993??? Okay, I guess it has been a while.

These days, Microsoft and Novell are as snugly as two peas in an iPod, and seem have been buddies since signing (in 2006) a collaboration agreement aimed at bridging the gap between open-source and proprietary software. This was followed in 2007 by opening the doors of a joint development facility where engineers from both companies began designing and testing software that would work with both Windows Server and SUSE Linux Enterprise -- interoperate, in other words. The tag-team days are a thing of the past. Instead, these days, there's a single, unified team of engineers with a shared vision.

To jump-start this shared vision, the team did things like cross-platform testing and training on each others' toolsets. For instance, the Novell team tested Windows on top of SUSE Linux Enterprise/Xen, while the Microsoft team tested Linux on top of Windows Server 2008 Hyper-V. According to Tom Hanrahan, director, Microsoft Open Source Technology Center, "We felt that it was critical to set up a collaborative environment dedicated to delivering not only better technical interoperability, but also tools that can help them transition to an enterprise-class Linux platform, which also works with their Windows infrastructure. Because of this, the testing and development work performed in the lab directly influences features and capabilities that are also being engineered into our respective products."

Take virtualization, for example. The lab's efforts mean that SLES 11 can be an optimized guest on Windows Server 2008 R2. The engineering optimized both products to work with one another so users can use and move Windows and SUSE Linux virtual machines from one physical host to another without any interruption in service or loss of performance.

Cool. Microsoft and Novell first. Next up, the Middle East.

-- Jonathan Erickson
jerickson@ddj.com

Even with a name out of Rocketship X-M or some other 1950's sci-fi flick, accelerometers are a very cool technology -- especially when implemented in mobile phones. Accelerometers detect and track the phone's movements by measuring the acceleration the phone experiences relative to freefall. They're what makes it possible, for instance, for the display to rotate from landscape to portrait modes. But there's nothing particularly exciting about that.

What is exciting, however, is the kind of application Sandip Agrawal and Ionut Constandache developed that uses the accelerometer in a unique way. The application, which they call "PhonePoint Pen," lets you write short messages in the air with your phone, then e-mail the messages. In other words, the application uses the phone's accelerometer to recognize human writing.

When all is said and one, this is simply another instance of gesture-based input, a topic that's getting a lot of attention these days. For instance, see Designing for Multi-Touch, Multi-User and Gesture-Based Systems; Gestures, Multi-Touch, and Microsoft Surface; and 3M Announces Multi-Touch Developer Kit. Then there are gesture-based systems being developed by companies such as eyeSight, and devices like the Logitech Air Mouse.

As cool as it is, the project isn't without its challenges. For one thing, accelerometers like that in the Nokia N95 mobile phone (which the researchers used) can measure linear movements, but not hand rotation. However, they did come up with a workaround, as described in their paper PhonePoint Pen: Using Mobile Phones to Write in Air. Other challenges included coping with background noise, the necessity to pause briefly between letters (which means that cursive writing is out), and the fact that each letter written must be large in size -- none of which are show stoppers, say the researchers, especially as better algorithms and accelerometers come into play.

-- Jonathan Erickson
jerickson@ddj.com

Sleep-mode aside, computers have assumed the dubious mantel of being the biggest consumers of electricity in the world. Sleep-mode kicks in when computers are inactive for a specified amount of time. On the flip side, awake-mode can use a lot of electricity, even if they're not being used. It was this either/or that led Yuvraj Agarwal, a PhD student at the University of California, San Diego, to seek some middle ground. And what he came up with is a "sleep-talking" power mode, which saves electricity like sleep mode, but still has some of the network-and-Internet-connected convenience of awake mode.

As described in Somniloquy: Augmenting Network Interfaces to Reduce PC Energy Usage, what Agarwal and his team ended up doing was building a USB-connected hardware/software system that lets a PC remain in sleep mode, while maintaining a network presence and running applications like Instant Messaging, VoIP, background web downloads, peer-to-peer file sharing networks, and remote access.

They call the system Somniloquy, which means "the act or habit of talking in one's sleep." Somniloquy has a low-power processor, operating system, 64 MB of DRAM, and 2 GB of flash. Somniloquy's low-power secondary processor functions at the PC's network interface. It runs an embedded operating system and impersonates the sleeping PC to other hosts on the network. Somniloquy will wake up the PC over the USB bus if necessary. For example, during a movie download, when the flash memory fills up, Somniloquy will wake up the PC and transfer the data. When the transfer is complete, it will go back to sleep mode and Somniloquy will again impersonate the computer on the network.

Does it work? According to Agarwal, Somniloquy-based systems consume 11 to 24 times less power than a PC in sleep mode, which could translate to energy savings of 60 to 80 percent. Agarwal believes that in the future, Somniloquy could be incorporated into the network interface card of new PCs, eliminating the need for the prototype's external USB plug-in hardware.

-- Jonathan Erickson
jerickson@ddj.com

Grok: Jump to: navigation, search.

If you're a Robert Heinlein fan, you know about grokking. But "System Grokking"? Maybe not so familiar.

The concept of system grokking starts with the assumption that software systems are becoming more and more complex with semantics that are hard to grasp, especially when it comes to legacy code. We can agree on that. We can also agree that our need to cut through the complexity is growing just as fast. And that's where System Grokking (or System Grokker, depending on who you talk to) comes into to play.

Developed by IBM researchers Dany Moshkovich, Maayan Goldstein, Avital Gutman, and Vadim Vasilov, System Grokker (or System Grokking) is a technology for understanding software systems. The researchers describe it as a "software architect assistance technology" that's supports incremental and iterative user-driven understanding through higher levels of abstraction. System Grokking enables semi-automatic discovery, manipulation, and visualization of groups of domain-specific software elements and the relationships between them, to represent high-level structural and behavioral abstractions.

In other words, System Grokking is a language-independent static analysis tool that builds models of existing software systems. Reverse modeling, kind of like reverse engineering, I suppose. To make this possible, the team designed a family of program-slicing algorithms, which they say are more accurate than current algorithms, especially for unstructured languages. They are also developing flexible and reliable code-motion refactoring algorithms.

Putting this another way, System Grokker makes a system more understandable by presenting it in higher and higher levels of abstraction and via different views. It also assists system validation by exposing problematic relationships and anti-patterns, and supports software evolution process by suggesting architectural improvements and simulating architectural changes.

-- Jonathan Erickson
jerickson@ddj.com

Okay, so I'm easily confused. At least I was for the first few minutes of Jamie Thomas's keynote at the Rational Software Conference when she introduced "MCIF," short for "Measured Capability Improvement Framework." When I hear the word "framework" along side the word "software," what usually comes to mind is something like the .NET Framework, Django Framework, or Java Struts Frameworks. My bad. MCIF isn't a code-based framework at all -- it's a conceptual framework and it is really interesting.

In a nutshell, Measured Capability Improvement Framework is a structured approach for providing software development teams with best practice guidance through measured and continuous process improvement. According to Thomas, Rational's vice president of development, the MCIF concept is based on four principles:

• The establishment of business and operational objectives.
• The priorization of practices.
• The acceleration of advanced tools adoption.
• Reporting, analyzing, and action.

In other words, MCIF provides a systematic approach to incrementally improving software. Optimized for agile teams and methods, MCIF provides an end-to-end framework that lets organizations measure results and manage projects to incrementally improve their software delivery capability. Depending on the specific goal, MCIF (which is built on a Jazz infrastructure) lets teams specify business and operational requirements and identify appropriate best practices using tools such as the Rational Asset Manager, Rational Requirements Composer, and others; although it should be noted that the concept is vendor neutral, allowing, say, requirement management systems from other vendors, such as Ravenflow. It is also process independent and can be used in conjunction with RPU, eXtreme Programming, Scrum, and other processes.

For the most part, best practices are gleaned from IBM's 10 years of experience helping teams incrementally improve their practice maturity to deliver business results. Parts also come from agile's transformation of numerous IBM internal projects. However, the best practice repository can include third-party practices as well.

To find out more about MCIF, check out this video and/or this webinar, both with IBM chief architect Per Kroll.

-- Jonathan Erickson
jerickson@ddj.com

Now here's a project I like because it involves issues that are meaningful to me -- mobile technologies, open source, and using technology for helping people. The Open Mobile Consortium is a collaboration of organizations that are creating open source software tools to serve the health, humanitarian, and development needs of the "bottom billion" -- the poorest and most-disenfranchised citizens of the world.

By some estimates, over the next couple of years, more than 1 billion people will log onto the Internet for the first time, and most of these log-ins will be via mobile devices (like cell phones) and located in emerging countries. It is also estimated that by 2010, one in three Africans will own a mobile phone. With almost 280 million subscribers in Africa alone, mobile phones are recognized as instruments of change in finance, agriculture, media and, development work. The Open Mobile Consortium was founded to develop and bring to scale free and open-source solutions that leverage the power and ubiquity of mobile phones.

The OMC's goal is to help organizations working in developing countries collaborate across disparate platforms and products, reduce redundancies, and create a mechanism for freely sharing technical tools, information, and approaches.

The OMC has already brought together a number of mobile technology tools for collaboration and sharing. These include, among others:

• CommCare, a mobile-phone based application that lets community health workers provide better, more efficient care and improve coordination of community health programs.
• Mobilisr, an open source enterprise class mobile messaging platform for NGOs around the world.
• Mesh4X, a platform for seamless cross-organizational information sharing between mobile devices, databases, desktop applications, and websites.

If you'd like to get involved with the OMC, click here.

-- Jonathan Erickson
jerickson@ddj.com

Schull, an assistant professor in MIT's Science, Technology, and Society program, studies how addictive sophisticated, high-tech gambling machines have become. With their high-intensity feedback, magnetic-stripe player cards instead of coins, and haptic touch-screen technology, Schull sees video poker machines as the "crack cocaine" of gambling machines, creating "a sense of transaction" that keeps gamblers gambling and gambling and ....

As Schull notes in her paper Digital Gambling: The Coincidence of Desire and Design, game designers have figured out how to keep gamblers to stay at machines by developing a nuanced sense of how to modulate technology to accommodate gamblers' play flow. For instance, according to Nathan Leland, designers add a dynamic play rate to video poker machines when they became too slow for experienced players: "The screen shows animated video hands that deal the cards. If you play slow, it deals slow. When you go so fast [imitates rapid pressing on buttons], the game detects it and adapts. Even then, some players were annoyed that it was too slow, and we had to speed it up. If a player goes fast enough, the cards pop up without any animation." Schull goes on to show that digitization lets programmers mathematically adjust payout tables or reward schedules for specific player profiles within a diverse market.

While I'm fascinated by all this, the thought of being unknowingly manipulated like this puts a chill down my spine. On the other hand, since I rarely visit a casino, it probably doesn't matter. But it does matter to a lot of people. Estimates are that most of the revenue from the U.S. gambling industry -- reported as $92.3 billion in 2007 by the American Gaming Association -- comes from machines like video poker.

Frankly I can't wait to read Schull's upcoming book Machine Zone: Technology and Compulsion in Las Vegas, which is due to be published by Princeton University Press later this year. And I'm going to track down the documentary film she made called BUFFET: All You Can Eat Las Vegas. You can bet on that.

-- Jonathan Erickson
jerickson@ddj.com

Everyone seems to be talking about cloud computing, but few people seem to know what it is. That's what InformationWeek's Nick Hoover discovered at Interop this week, at least according to his report Vendors Still Confused With Cloud Computing Definitions. Even vendors like IBM, HP, and SAP -- who should know what it is -- seemed unsure. What would clear up the confusion? One thing would be standards.

One approach to standards for cloud computing is what Andrzej Goscinski, a professor at Deakin University is compiling. He is working on a framework for building infrastructures that are more accessible, reliable, efficient, and yes, understandable. Goscinski's approach builds on his earlier work involving a Resources Via Web Instances (RVWI) framework, which bundles state information of a web service into its WSDL. Goscinski and Michael Brock describe RVWI in their paper State Aware WSDL: The Resources Via Web Instances Framework.

Then there are the folks at NIST, the "National Institute of Standards and Technology," who ought to know a thing or two about defining standards. Computer scientists at NIST, in collaboration with industry and government, are producing a special publication that covers cloud architectures, economics, security, and deployment strategies. To get thing started, they've put together a working definition of cloud computing:

Cloud computing is a pay-per-use model for enabling available, convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, services) that can be rapidly provisioned and released with minimal management effort or service provider interaction. This cloud model promotes availability and is comprised of five key characteristics, three delivery models, and four deployment models.

These aren't the only candidates, however. There's the Open Cloud Consortium with its proposals, and the Distributed Management Task Force with its. And that's just for starters.

So don't get in a hurry. It won't be long before we have plenty of cloud computing standards. And that's when things will get really confusing.

-- Jonathan Erickson
jerickson@ddj.com

You can blame it on IBM. Or maybe Garry Kasparov. I suppose it really doesn't matter. But what does matter is that after Deep Blue beat Kasparov at chess, the game of Go seems to have replaced chess as *the* testbed for AI research. Go is one of the few board games that humans can still whip computers at -- for the time being, anyway. New algorithms and more powerful computers may be changing that, too.

Thanks to Monte-Carlo Tree Search algorithms, developed in 2006 by Remi Coulom and described in his paper Efficient Selectivity and Backup Operators in Monte-Carlo Tree Search, AI systems have a better chance at turning the tables on human players.

For instance, at the Taiwan Open 2009 event held a few weeks ago, the Dutch national supercomputer Huygens running an application called "MoGo TITAN" (developed by INRIA France and Maastricht University) defeated two human Go professionals in an official match. This is the second victory for Huygens playing Go against professional players. The first Huygens victory was last year at the 24th Annual US Go Congress competition, held in Portland, Oregon.

Huygens is an IBM Power 575 Hydro-Cluster model, which uses a unique water-cooling system that is 33 percent more efficient than traditional air-cooled systems. Heat is removed from the electronics by water-chilled copper plates mounted in direct contact with each Power6 microprocessor. Huygens is the first Power 575 Hydro-Cluster to run Linux. It is planned to achieve a peak speed of 60 trillion calculations per second (Teraflop/s). The system has 3328 4.7-GHz processor cores, a total memory capacity of 15616 GB, and 972 TB disk capacity.

According to researchers, MoGo illustrates three technological advances:

• The so-called Bandits Manchots ("One-Armed Bandits") algorithms made it possible (partially) to explore the space of possible moves. They thereby revolutionized the world of planning in an uncertain universe.
• The evaluation of positions is based on Monte-Carlo algorithms, simulating the behavior of a low-level stochastic player but without any bias.
• Parallelism on a large scale served to provide the computational power required for a Monte-Carlo evaluation to give sufficiently precise results.

Moreover, these experts explain, the game of Go poses scientific problems that are difficult to solve. Even if the game is determinist in theory, meaning that it can be programmed without any special difficulties, it would take an astronomical amount of computation time to obtain the solution, say researchers. However, in this case, the parallel algorithms use computational grids to solve problems concerning the sequencing of possibilities and of planning. The sequencing of possibilities amounts, for example, to thousands of millions, and the objective is to find the most efficient order in this mass of possibilities. The bottom line, say researchers at INRIA, is that the game of Go is thus a good model for developing and testing new techniques for solving this type of sequencing problem.

-- Jonathan Erickson
jerickson@ddj.com