sgerakin@hawk.ulowell.edu (Steve Gerakines) (09/10/90)
Well in my bliss of successfully upgrading to Minix-ST 1.5.10, I just realized that there is no math lib for the C compiler. To make a long story short, I dug up a port of P.S. Housels Minix-PC floating point package, and it seems just dandy. The only thing I didn't like about it was using a silly "fcc" batch file to "convince" cc to allow floats. I came up with a real simple solution that's transparent to cc, doesn't require patching binaries (ick!), and doesn't require finding a new "cc" program. Basically all I did was write a program called "cg", and rename the original "cg" to "cg.real". The "cg" program I wrote just plays with the command arguments, and forks off two processes for "fpp" and then "cg.real" right afterwards. Brute force, but it works. :-) This little fix makes the order nice and clean: /usr/lib/cem prog.c prog.k /usr/lib/opt prog.k prog.m /usr/lib/cg prog.m prog.s the usual parameters, /usr/lib/fpp prog.m prog.n then play switcheroo to fool fpp /usr/lib/cg.real prog.n prog.s and cg.real /usr/lib/as prog.s prog.o Anyway, I thought it was pretty sneaky. Anyone else come across this before? I'd like to hear some other suggestions. Now maybe I can get my mtv tracing. :-) -Steve Gerakines ---------------------------------------------------------------------------- | Usenet: sgerakin@hawk.ulowell.edu | SteveNet: GENESIS:Steve2 | | UUCP: ...!harvard!swan!sgerakin@hawk | "My kingdom for a smoke!!!" | ----------------------------------------------------------------------------