[rescue] systems with user modifiable microcode.

[Skeezics Boondoggle]

On Wed, 2 Oct 2002 17:13:05 -0400, Joshua D Boyd wrote:

> > I thought the sparc tagged data was used in some older LISP
> > implementations?
> 
> I've seen discussion of it, but never any implementations.  One issue is
> that it makes it a bugger to inter-operate with existing C code since
> the pointers will be wrong, and the numbers funky.  Not that it couldn't
> be made to work, I just haven't seen any examples of it being done that
> way. 

the symbolics lisp machines had tagged instruction sets... they did the
same thing on the perq for "spice lisp" too -- and you could dynamically
load it as needed!  the accent kernel, disk and ethernet drivers, and
c/pascal/fortran "q-code" interpreter ("q codes" were the higher-level
instruction set; think ucsd p-system) were generally loaded into the wcs
at boot time.  when you fired up the lisp environment, the loader would
actually page in the lisp instruction set if it wasn't already loaded!  
and it all fit in 16k x 48 bits of control store...

another silly example of how cool it is to have a "soft" instruction set
was the "ROT13 opcode", written in response to a challenge to write the
shortest program to rot13 a string.  i bet google would turn up the
source.  not sure if anyone ever tried to run it, though - i should fire
up the old perq and test it out. :-)

why nobody ever did smalltalk on the perq i'll never know.  i was thinking
it probably wouldn't be too hard to write an x86 emulator, or maybe even
an m68k emulator - turn my perq into a 200-lb, 720-watt, 5.8mhz "fat mac".  
oooh.

-- skeez