Re: Hurd [was:M$ licenses Unix]
Mike M wrote:
In the introduction the term "natural abstraction" was used. It started the
wheels turning in my head and I could read no further. Are there
abstractions in nature? I thought nature was the uncontrived - the
unabstract. I'll ponder the question as I compile a kernel - I'll need make
mrproper for this one - and 'M' on all the config options - and an old
Pentium with not much memory.
I'm not exactly sure what they mean by "natural". One of the issues in abstract
state machines (ASMs) is whether or not they truly represent the intuitive
(natural) concept of algorithm. that might be it.
I'll check on the Hurd, Mach and L4 lists for more on this. If anyone's
interested, I encourage you to write down some abstract operations that a
microkernel would have to implement. Maybe they're proceesses, maybe only
threads, I don't know. But the important thing is to find a minimal set that'll
support all the other operations. Then you can squeeze them into an ASM form;
that's the tricky step. ASMs are like FSMs. The difference is that a state in an
ASM is analyzed into functions, whereas an FSM state has no composition, only an
identifying number or string. Determining those functions and the transition
rules that change one state into another is where the serious work lies.
Quoting from the ASM website: "If we stick to one abstract level
(abstracting from low-level details and being oblivious to a possible
higher-level picture) and if the states of the algorithm reflect all the
pertinent information, then a particular small instruction set suffices in all
cases." So far so good, but then when it's time to stuff it into an ASM chassis
and run it through something like Xasm, one needs serious help from one's
friendly logician neighbors, assuming they haven't all moved to Redmond with
Yuri and bill ...
Enjoy. :-)
-- TT
Reply to: