Programming projects I did for fun

• While I was a student I attempted to write an an emulator for my HP-67 pocket calculator (of which I was immensely proud) in FORTRAN'66 (of which I was not). Later, when Apples II became available, I redid this in 6502 assembly code. The project was partially successful: features emulated included full programmability, and arithmetic functions up to the complexity of square root extraction (with exact emulation of floating point BCD arithmetic, to the point of reproducing precisely the same round-off errors); transcendental functions like the logarithm however were never implemented.
• On a stand-alone `intelligent terminal' from DEC, with an LSI-11 processor, a DEC-2 tape drive as only mass storage device, and running the RT11 operating system (this was at the time that Apple II was the only available microcomputer), I wrote a FORTH-like kernel in PDP11 assembly language (given these items and the next you may start to think that assembly is my favourite language, but it is not; in this case it was about the only language available except BASIC). I called the system Interactive Threaded Code Assembler, and it allowed interactive assembly and execution of both ordinary machine code and threaded code. I actually used the system for such things as solving certain geometric puzzles, which would have been hard to do by any other means on that machine. The project was discontinued because I moved out of the house of the owner of this curious machine, and I never saw another one since.
• I wrote an interpreter for a simple and mostly functional language, based on the basic ideas of LISP, but more decent in syntax and semantics; it allows (and heavily uses) higher order functions, but is not statically typed. It is extremely non-portable, as it is entirely written in 68000 assembly code (but with parsing tables generated by yacc), and runs only on Atari ST computers. Some interesting features are its speed, given the platform and its interpreted nature (it computes 1000! in about 5.5 seconds, where an internal routine takes another 3 seconds to convert the result to decimal for output), its compactness (the executable is 16 Kb; the earliest version took only 5 Kb, which already included a garbage collector) and the fact that it has an implementation of unbounded integer arithmetic that uses a 2's complement representation. To my own surprise, this language has turned out to be my language of choice for several experiments with combinatorial algorithms, whose basic definition is easy to implement, but where I need very flexible access in applying the algorithms.
• As a `first exercise' in compiler construction, I started writing a full language Algol68 compiler, implemented in C. The project is currently stalled with the analysis phase almost complete (all errors as defined by the report will be caught, except those involving transiency of names, which was deliberately postponed). Before I start the synthesis phase, I shall need to study code generation; in the mean time I shall complete conversion of the program into a literate one (using CWEBx), which may also involve writing an extension to CWEBx to conveniently handle yacc grammar specifications.
• My only serious program with a graphical interface is one to display and manipulate `pictures', in the mathematical sense as defined by Zelevinsky; it focusses on the operation of glissement, as described in my pictures paper. Actually, the graphical interface using Xlib was originally written by a visiting student, James Faux. This program was not written just for fun, but also because I really needed it for experiments with the algorithms defined in abovementioned paper. Although the program works, and is entirely written in CWEBx, the sources still leave much to be desired, and I plan to rewrite and possibly extend the program some time in the future.