[FoRK] ridiculous project

rst at ai.mit.edu rst at ai.mit.edu
Mon Jul 8 06:53:53 PDT 2013


Joseph S. Barrera III writes:
 > I started reading again about the EDSAC and I've been wondering if you 
 > could possibly fit a BASIC interpreter in 1024 words of 17 (!) bits 
 > each, and leave room enough for a BASIC program. It would be hard as 
 > even TinyBasic used 2 or 3 KB. Actually, maybe LISP would make more 
 > sense? The first step of course is to write an EDSAC emulator. There are 
 > of course a few but I feel a need to write one in JavaScript so you can 
 > run all this stuff in any browser.

Well, there are a few interpreters described in the Annals of the 
History of Computing article on the EDSAC here, if you can get
access:

  http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=728229

However, these didn't have syntax nearly as free as BASIC; they
were deliberately designed to interpret "pseudoinstructions" similar
enough to the real ones that the assembler in the EDSAC's boot ROM
(the "initial orders") could be used to conveniently.  

Those "initial orders", by the way, were an example of cramming a lot
of functionality into very little space: they provided much of the
functionality of a modern linker (in particular, relocating code
segments) in 41 instructions.  But there was one severe omission from
the syntax: they didn't allow the entry of full-word numeric
constants, so numeric constants were entered as "pseudo-orders" ---
instructions which were not meant to be executed, but happened to have
the same bit pattern as the number desired.

Infix syntax would be a real stretch...

rst


More information about the FoRK mailing list