MWC (was: Re: How the Mac was born... supposedly...)

Mark Davidson wrote: Much of the compiler work, at least when I got there, was done by Charles Fiterman, who was the kind of genius that could whip up a C compiler for a new platform in about 2 weeks -- and a peephole optimizer in another week -- but if you gave him a task that was way too easy for him, he would do it almost without thinking and there would be dumb obscure bugs caused by things like incorrect order of operations, misplaced semicolons, etc. I fondly remember him as a bit eccentric and very smart (in fact, most of the people who worked at MWC were incredibly smart, and I'm not just saying this because I worked there, it's the honest truth). Charles later left Mark Williams Company to form Geodesic (sp?) Systems to build and market a garbage collection library for C. I still have boxes of disks for the MWC C compiler for Atari ST. :-) No Let's C though, although I remember Let's C being advertised in Byte and Dr. Dobbs' before I worked there. The thing about MWC compilers is that they were *excellent* at optimizing for size. Not speed, but size. This led to some obscure tricks in the output bytecode that made you scratch your head in a disassembler, but it was worth it on systems with such limited memory. While we eventually ported gcc to the 386 version of Coherent, I still held great respect for the MWC compiler (even though it was K&R, not ANSI) because of the ludicrously small executables it produced. My numbers may be off, but I seem to remember that a statically-linked binary of GNU TAR as produced by gcc at that time was ~256K, while the same program statically-liked from the MWC compiler was ~50K. -- Jim Leonard (trixter(a)oldskool.org) http://www.oldskool.org/ Want to help an ambitious games project? http://www.mobygames.com/ Or check out some trippy MindCandy at http://www.mindcandydvd.com/