Presentations
A Basic Object System Using Macros [pdf] [zip]
Presented by
Grant Rettke on May 16, 2008
Grant is a strong proponent of using "toy projects" to learn fundamental concepts of computer science. In this particular case, Grant wanted to better understand objects and closures, so he iterated through several versions of a simple object system implementation using closures, which was further refined with the use of hygenic and unhygenic macros. In the process, he increased his exposure to language features and concepts. His final system included methods and variables, encapsulation, message passing, lexical scope, and name collision detection. He approached it as a code generation problem - write the code you want, then abstract out the essence using macros and syntax extension.
Some great closing thoughts that inspired rigorous discussion:
- Toy projects are a great learning experience.
- Lisp is a great vehicle for exploring ideas, which you can then use in other languages.
- Don't become a language slave, rather embrace the best ideas from all languages.
- Put in the time and effort to improve your study habits.
- Lips introduces powerful ideas that are hard to find in other languages.
For the detailed summary of this meeting, please see Peter's
blog post.
Combined Object-Lambda Architectures [pdf]
Presented by
John Quigley on April 18, 2008
Our group's inagural presentation was concerning the work being done by the
Viewpoints Research Institure (VPRI). The most recognizable name associated with this organization is Alan Kay, the inventor of object-oriented programming, Smalltalk and many other core technologies. John Quigley spoke about the paper,
"Making COLAs with Pepsi and Coke," (pdf) by Ian Piumarta. He led discussion of a combined/late-bound object-lambda architecture, it's prupose and meaning, and how such a system might be bootstrapped.
The COLA technology is being designed as part of Kay's
Fundamentals of New Computing (FONC) project at VPRI. The aim of the project is to create a complete computing system in 20,000 lines of code. This system would encompass everything from the hardware to the UI, including both design and implementation. Having such a compact system would be a useful exploration and learning tool, so every part of it can be inspected and manipulated. They're currently 18 months into a 5 year project, but they already have made remarkable progress. Read the
full meeting recap at Peter Christensen's blog.
