alex@umbc3.UMD.EDU (Alex S. Crain) (01/19/88)
In article <3285@ems.Ems.MN.ORG> mark@ems.Ems.MN.ORG (Mark H. Colburn) writes: >> Since KCL is so obnixous with regard to image space, I am thinking >>harder about MIT Scheme as I hear it runs a smaller interpreter (I have never >>used it). Has anyone ever used Scheme on an AT&T machine? email & I'll >>summerize if i get a response. > > I have tried to bring up Scheme version 3.5 (?) on my 3b1 at home. >Unfortunatley, the d*mn thing will not compile due to the way that it is >written. It will always fail with a 'too many defines' error message. I have been contemplating replacing the stock /lib/cpp with the GNU cpp that comes with emacs and gcc. I have it up and running for gcc, and it supports just about every option i've ever heard of, and quite a few that I haven't. Right now I'm just making sure it's bug free, before I make it the default. It should solve the above problem, though. I also have gcc (The GNU C Compiler) up and running. It works, although there are bugs in the optimizer that cause cc1 to crash on certain files. My copy of emacs compiled without qualms (some files without optimization) to a smaller (5k, even without -O on 1/3 of the files) executable, as did gcc itself, and a few others. (Funny, the only files that don't optimize are GNU software :-)). The compiler works well (no known bugs) without -O. The diffs are about 140K for version 1.17 of gcc. Earlier versions of the compiler will require Ed Hepler's diffs to get it to compile on system 5. Email if you want some. I will continue to work on the bug(s) in -O, soon as I finish my paying jobs :-), and would appreciate bug reports. Much thanks to Ed Hepler for his help getting me started with this project. -- :alex. alex@umbc3.umd.edu