Qi (programming Language) - L21 Project

L21 Project

Qi was developed as part of the L21 project, which aims to modernise Lisp to meet the challenges of twenty-first century computing. In his lecture, Tarver outlined a series of current challenges to Lisp which he argued were damaging to the wider use of the language. Specifically he identified Common Lisp's lack of pattern-matching, inconsistency with respect to lambda calculus theory (partial applications being missing), procedural contamination and lack of static typing. These deficiencies, Tarver argued, had led to a general abandonment of Lisp as a teaching vehicle at university level with a concomitant lack of graduating Lisp programmers. Tarver characterised this lack of support for teaching Lisp at university as leading to a 'classic vicious cycle', whereby the small number of graduating students fluent in Lisp encouraged a transition away from using Lisp programmers which, in turn, fueled the perception that Lisp was a dying language and fed the decline in the teaching of Lisp.

In the same lecture, Tarver suggested that this problem could either be tackled at the industry or the university end, but that tackling the problem at the industry end required a champion with large amounts of capital to invest in Lisp. Tarver instead proposed to tackle the problem at the university end, by modernising Common Lisp in such a way to make it 'future proof' for at least 25 years. His characterisation of an adequate modernisation of Lisp was summarised in ten requirements which Qi was designed to meet. The solution should:

1. be Lisp compatible—as the most widely used dialect of Lisp, the solution should be written in Common Lisp and run under Common Lisp.
2. be free for non commercial and educational use.
3. be compact—programs should not be any longer than the same programs written in Common Lisp
4. be simple to learn -- Semantics and syntax should be learnable by a child.
5. be efficient—The solution should generate programs which are at least as fast as their hand-coded Lisp equivalents. In practice, Qi programs have proven to be often faster.
6. have the characteristics of a modern functional programming language. By these Dr. Tarver includes pattern-directed list handling, static typing, currying, and partial applications.
7. not be simply a clone of Haskell or ML—this is part of what constitutes 'future proofing' the solution.
8. be computationally adequate—meaning that the language is 'adequate for the needs of the programmers of the day'. Dr Tarver characterises 'computational adequacy' as 'a large, vague and important notion' and believes that the extension of this concept changes through time. Common Lisp he characterises as 'computationally inadequate' due to the lack of specification for features such as foreign languages interface and graphics. Tarver believes that Qi itself still needs to achieve this goal through the development of standard libraries for graphics, web interfaces and foreign language handling.
9. be metaprogrammable and customizable—in that the user should be free to program the syntax of the language. To achieve this Qi has both M-expression and S-expression level syntax for all its syntax, as well as an "eval" function that maps M-expressions to S-expressions; which can make it easier to do metaprogramming in Qi than in Lisp. In terms of customization Qi has an explicit "sugar" keyword to allow the user to program the Qi reader to accept custom syntax.
10. be well documented and theoretically secure—Qi is fully documented, with formal correctness proofs and has a canonical textbook which is also online.