Sunday, September 13, 2009

Davinci: Open Source Engineering Tools (Part 1)

Tonight I want to talk with you about something I have struggled with for some time to figure out. This idea has been with me for close to if not over a year, and yet until recently I was finding it very hard to describe to others. I even posted about it once in my old ideas blog, but even as I wrote that version I felt frustrated by my lack of clarity on the topic. This is my second opportunity to do it justice.

I am, as you may have gathered, a advocate of tools. While it is becoming clearer by the day that humans are not the only animals with the ability to conceive and utilize tools to achieve our aims, it is one of our defining characteristics which has allowed us to thrive as a species. This is one of the reasons I am such a rabid advocate of open source. It calls on the better angels of human nature to facilitate collaboration in the search of better tools, accessible to all. Yet despite all of the wonders open source has provided the world of software, I would hazard that I am one of many who feel that open source must expand beyond the realm of software to deliver its best gifts to humanity. Open source must breach the divide and become a tool for the innovation of corporeal inventions. Several attempts at this have already been made, or are under way. However, their dream will never reach full fruition without confronting a basic reality: Advocates of open source hardware lack the equivalent tools that their software compatriots take for granted. Without these tools, open source hardware cannot achieve the same success that open source software has enjoyed.

When a coder sits down at a computer to write a program, they have all the tools necessary for the act of creation and collaboration at their fingertips. Code can be written in a free text editor or software editing application. That code can be compiled, for free, by a compiler residing on the very same computer. The coder can test the fruits of their labor for free and in most cases, without fear of harming themselves, their computer, or their work. Now to be fair all of this functionality can be mimicked by a engineer, utilizing a CAD/CAE program. A engineer can design a piece of machinery and test its basic functionality, safely and (depending on the software) fairly cheaply. In this regard, the two systems are relatively similar. However, the differences begin to emerge when collaboration, a essential ingredient of any open source project, comes into the picture.

The power of open source derives from the ability of a individual coder to contribute a relatively small amount of work which can easily be merged into a larger, more complex project. The core of open source is the acknowledgment that not everyone is Superman, and that many people contributing just a little can add up to something greater than its part. A coder working on a open source project can easily download all or part of the larger project, make changes, compile the entire project and test it. Adding the work of others to a existing program is also relatively painless, requiring only a few lines of code to instruct the program how to access the new code and to call its functions when needed. Indeed, the command #include and its kin are one of the most powerful commands from the perspective of open source. They are powerful because they allow for a single coder to quickly add the work of another coder, making collaboration not only easy, but in many cases easier than working completely alone. This is where the design of real things runs into trouble. A team of committed, determined engineers looking to create a large open source design, they will quickly realize that while they can all design individual pieces and test them all individually, there is practically no way to test the entire system without building a physical prototype and testing its performance. While this might be a acceptable solution for something small, simple, and cheap to build, it becomes a serious problem for larger projects. The average contributor is most likely a person of modest means, and probably could not afford to build a functioning prototype of something large, like a car, building, satellite, or playground. There should be a tool that empowers the average contributor to the same level that a simple compiler empowers a coder. Such a tool should allow a contributor to build, edit, test and share large complex projects. That is what I shall attempt to describe.

Initially, I envisioned such a tool as a monolithic piece of software. This one program would handle all functionality, from the design of individual elements, all the way to the testing of large complexity projects. I based this initial notion off the analogy of a software development environment, where code writing, project organization, and testing functionality were all part of the same program. Understandably, this system became very hard to describe, as I tried to describe a family of functionality while retaining the notion of a singular program. It wasn't until recently that I realized what I was really looking for was a close-knit ecosystem of smaller, function specific programs. Once broken down into functions, the system is suddenly much easier to conceptualize, and hopefully easier to describe.

Imagine my embarrassment...

Continued in Part 2


Post a Comment