palevich.NBS-VMS@BRL@sri-unix (07/16/82)
I've ported Ron Cain's Small-C Compiler to the 6502, (to the Atari 800 Home Computer) and I remember that the benchmarks (like the High-Level-Language Benchmark out of the Sept. '81 Byte) placed its speed at six times faster than Atari Basic, or about half the speed of the Atari Pascal compiler. The actual time was 440 seconds, I think, which isn't very fast. (see p. 192 of the September 1981 Byte for other times) The big problem is the 6502 architecture, which makes it impossible to write compilers. My implementation is yet another pseudo-code interpreter. Ah well, "the amazing thing is not that it compiles quickly, but that it compiles at all." On a related topic, my attempt to port Ron Cain's C to the TRS-16 was foiled by the pathalogical stupidity of the TRS-DOS16 operating system. For reasons known only to the innermost circle of Tandy systems "programmers", the 68000 assembler & editor use a new and incompatible file format, and no format conversion program exists. A warning to people bringing Small-C up under VMS -- the current version of the VAX VMS C compiler can't store more than 7 bits of information in a character (not even as negative numbers), so you have to re-code all lines using "offset".