CompuServe Messages

Open Letter to CBM

    12-Oct-91 02:02:14
Fm: Don Curtis/SYSOP 76703,4321
To: Harold W. Grigdesby 76304,3030
Hal, Sure…ABI compliance is a major plus. It allows an ABI program written for ANY 68xxx SVR4 machine to run on any other 68xxx SVR4 machine AS IS. In other words, true shrink-wrap software for UNIX. As to why the 3000's UNIX (it's not called AMIX any more…or last I heard, it wasn't called AMIX any more) wasn't fully optimized for the hardware. I *suspect* there's multiple reasons. A. It was a first version, thus the idea would be to get it running, then find the weak points and optimize in a later version. That's NOT to say there weren't some optimizations…but it's obvious it wasn't fully optimized…else why would the 2.0 version of UNIX run faster? B. SVR4 itself was in flux. My memory is that it was November of 90 when UNIX, Int'l made their first release of code available. C. The quality of the assemblers, compilers wasn't as good. D. Hand optimization takes a long time (relatively), thus early versions may have been optimized by the compiler/assembler while later versions may have been hand optimized. The above is all GUESSES, I have/had no special knowledge of what went on inside Commodore. Don p.s. BTW…beta versions of SVR4 were available before Nov of 90, but the final version wasn't available from UNIX Int'l until then. Again..that's my memory, I could be wrong.