| Mirror on informatimago.com
| Mirror on free.fr
|
Copyright (c) 2003-2006 by The BioBike Team (contact Jeff Shrager for more information)
The question that we are most frequently asked by engineers and computer scientists is: "Why did you base BioBike on Lisp, instead of some 'modern' programming language, like C/C++, Java, PERL, or Python; Isn't Lisp an archaic, dead, and anyway impossibly complex programming language?"
First, Lisp is far from archaic or dead! Although it is one of the oldest programming languages (invented in the early 1960s, along with Fortran!), it is still widely used in the Artificial Intelligence community, and is the first programming language taught in intro computer science courses at many top universities, including MIT and Berkeley. (Actually, what's taught is usually Scheme, a dialect of Lisp; Common Lisp can be thought of as an industrial form of Scheme.)
Lisp is also the only programming language with certain very useful features, such as true macros, that are critical when you are inventing a new programming language, like BioBike, because they enable you to extend Lisp's syntax1. Lisp is the most natural language in which to write complex 'symbolic reasoning' programs, which are the heart of BioBike's novel capabilities. Symbolic reasoning involves manipulating conceptual objects, such as 'genes', 'ontologies', 'pathway models', and the like, and Lisp was built specifically for this sort of manipulation (although not specifically for biology at the time). This is why it has remained popular in the most advanced computing circles, esp. in Artificial Intelligence.
As for being impossibly complex, it's not any more complex than any other programming language. (Those who don't know Chinese think it's hard too, but the Chinese don't!) Many other 'modern' programming languages have been based directly on Lisp; Python, for example, is just a near approximation to Lisp. (See Peter Norvig's comparison here.) One of us (Shrager) teaches Lisp to both programmers and non-programmers regularly. Lisp is often taught to non-programmers because it has very simple syntax (requiring only the balancing of parentheses), and deals naturally with complex symbolic computations (as above), so musicians, writers, and others who want to write programs that deal with complex structured knowledge find Lisp easy and powerful. We think that biologists will as well.
As for C/C++/Java, those languages are of a very different sort than Lisp and Python (and to a large extent PERL). Lisp, et al., are essentially interactive programming languages; You write small concise programs that do small concise things, and test them in small concise ways. C and friends are designed for writing large, industrial-strength programs, and have lots of features -- such as required type declarations -- that make them run FAST, but which at the same time make them difficult to use for small, concise applications. (Lisp, by the way, is no slouch in the area of speed! It has decades of excellent compiler development underneath that makes well-written Lisp code run at nearly the speed of C (better than half as fast, in many cases). Of course, you can write a slow program in any programming language! Also, as pattern matching is a big thing these days, it's interesting to note that a PERL-compatible pattern matcher written entirely in Lisp is as fast as, and in some cases faster than, the PERL pattern matcher itself! Code and comparisons here.) We use this pattern matcher.
Bottom line: Why Lisp? Because it's the right language for the job!
Other essays along similar lines:
Notes:
1. Python, which is more-or-less lisp w/o the parens, is missing several critical features to do what we do in BioBike, esp. true macros, which enable you to manipulate expressions, create sub langauages, and engage in applicative programming. Someday the Python crowd will realize that uniform syntax is a good thing, and will re-adopt parens (or something like them), and then Python will just be Lisp again, and much the better for it!