HBO043%DJUKFA11.BITNET@cunyvm.cuny.edu (Christoph van Wuellen) (12/09/90)
I plan to add modules to c68 to get a 386 C-compiler. It will be 32-bit ints only. Since many other guys told me they planned exactly that, will they please make progress reports ( I heard nothing ). I will post it (if it succeeds) with a simple-minded code generator, no register variables ( even the full version cannot support more than two, in edi and esi), and a subminimal peephole optimizer. The netlanders can improve it later. HANDICAPS: I do not like INTEL chips very much and I have no literature about this, but I have access to a Sun386i workstation with 386 processor, the Sun C-compiler and the GCC1.37 C compiler for the 386 as a reference. I would rather make a compiler for the T800 or some RISC chip, but I have no machine to play with. Since the 386 evolve to be the second platform for "real" MINIX (besides 680X0, of course), it may be a worthy project. I intend to start the ball rolling and make you work on it, too. I expect more resonance than for c68, which is accepted by few people since it cannot win against ACK on its home turf. C.v.W.
agm@stl.stc.co.uk (Andrew G. Minter) (12/10/90)
In the referenced article HBO043%DJUKFA11.BITNET@cunyvm.cuny.edu (Christoph van Wuellen) writes: >I plan to add modules to c68 to get a 386 C-compiler. >It will be 32-bit ints only. Brilliant! I'm dying to have a C compiler with full source that I can mess about with. I wonder how it will compare with BCC, I don't have any idea what BCC is based on. GCC is quite nice, but it's so huge and slow. Andrew