File : README SCCS : "@(#)99/09/10 README 1.1" Author : Richard A. O'Keefe Purpose: Explain pcfpcmp Programming Contest Floating Point problems need a Programming Contest Floating Point Comparison and Monitoring Program: pcfpcmp. The documentation for pfcpcmp is in the source code, pcfpcmp.c. To install it, do cc -Dunix -o pcfpcmp pcfpcmp.c on a UNIX system. It does not depend on anything outside the ISO C standard. To compile it on any other system, use your compiler's options to set strict "ANSI" conformance. Macintosh users: this will need some kind of GUI wrapper around it. There are only four actions needed: Choose Model file Choose Instance file Run check Quit so it should be easy. To make it simpler, the code has been written so that main() only decodes the arguments, while check() does everything else. It uses setjmp/longjmp to catch errors. Instead of writing to stdout, my_exit() should put up an alert box. my_exit(EX_SOFTWARE) should quit the program after the alert is displayed. Don't bother about optimisation or switching off assertions. Programming Contest Floating Point Tagging and Adaption Grinder: pcfptag lets you use pcfpcmp for model answers that do not contain floating point numbers, and for the first stage of developing ones that do. You write a model answer with no excess blank lines at the end and no excess blanks at the end of lines, and no special care about reserved characters. Then pcfptag model or pcfptag draft >model adds the necessary magic characters to allow extra blank lines at the end of an answer and extra blanks at the end of answer lines. Floating point markup you will have to add yourself, but pcfptag rawmodel | pcfpcmp - submission compares a submission against the raw model. The same notes about compilation apply. For a Macintosh, or any other GUI interface, there are just four actions: Choose input file Choose output file Run filter Quit If someone wants to program that, I'd be delighted to have it.