🚀 go-pugleaf

dan@dna.neusc.bcm.tmc.edu wrote:
>In <41irqv$t83@usenet.ucs.indiana.edu>, sgoehrin@copper.ucs.indiana.edu (scott goehring) writes:
>>steven bourne's memory management scheme for bourne shell was to just
>>write the memory and, if you get a sigsegv, sbrk some more memory and
>>restart.  works fine on a PDP-11.  not so good on other architectures.
>
>	Just wondering what is the problem with this strategy. Admittedly it is
>somewhat unconventional, but it is just implementing a form of user-level VM.
>Instead of realloc()ing, which I find to be very messy and inefficient, for a
>growing contiguous data structure, Bourne's method seems to be what you want.
>Don't know what examples you are thinking of that make this not good.
>Interdata perhaps. But a decent architecture should allow growing memory
>requirements and growing stacks as well.
>

The problem is that no one -- including Dec -- has correctly followed the
example which the PDP-11 provided.  (And all have tried.)  I'm not
inclined to consider the strategy of relying on faulting to implement
features unconventional -- but I'm also used to the PDP-11's way of doing
things.

Someone bothered to do a little Group Theory when designing the 11, and
it seems that no one has bothered to try the same technique ever again.
(Actually, Motorola has paid a bit of attention to the basic math...)

-- 
Sphere.

Find a Linux/GNU Group for you: http://www.tiac.net/users/williams/lugnuts/
Buy Free UNIX!
Microsoft is prohibited from examining any packet containing data
originating on any machine which I am using. (Not that I can stop them.)