>On Mon, 11 Feb 2002, John Hall wrote: > >> In my opinion it is inevitable that a substantial amount must be >> written in assembler (assuming, of course, that you think it is >> appropriate to write *ANY* of it in a high-level language). The >> classic example is the memory test, where the only storage you can >> rely on is the processors register file... > >The challenge here is that having a portable OS, which could at the >least be run on m68k and ARM architectures, would require distinctly >different initialisation and that if parts of the OS itself were >written in assembler, there would have to be two separate yet >synchronised code branches.
<snip> >My ARM board will happily run uQLx, what would be nice would be to >open up QDOS+ to something like ARM or MIPS, so the OS isn't so tied >to the hardware... This, of course, is the crux of the matter - can SMSQDOS be ported to another processor/language and retain its distinctive character? And if it can't, what's the point? As someone (Peter, I think) pointed out, there are literally dozens of embedded/real-time operating systems out there, most of them *DESIGNED* to be ported to different platforms... FWIW, I think the closest we could get would be a *SMALL* machine code kernel that presented an SMSQDOS-like register-based interface to applications and managed a set of installable modules (perhaps we could come up with a mysterious name for them - how about 'Things'?) that provided most of the OS facilities (e.g. file systems) and could be written in 'C' or whatever. Then "all" we would have to do is rewrite SBASIC, the PE etc. in the language of choice, come up with an elegant scheme for supporting 68K application code on the chosen processor and Bobs your uncle :-) Regards John -- [EMAIL PROTECTED]
