Re: Forth

Andrew Dalke <dalke-DxsMES/F/[email protected]>:
> I just waste^H^H^H^H^Hspent the last 5 hours or so implementing
> a stack interpreter for the Turing Machine Instruction Set.  It
> uses unary numbers, so
>    0 ==
>    1 == 1
>    2 == 11
>    5 == 11111
>     etc.
> It does not understand negative numbers.

I wasted a bunch of time writing a macro assembler for Turing
machines.  I wanted to write a binary multiplication program, but the
binary addition program was so painful that I wrote the macro
assembler instead to help me.  You get to write things like

        !alphabet _ 01  
        
        !macro scan right to $sym then $newstate
        start   $sym backup    $sym R
        start  ^$sym start     *    R
        backup *     $newstate *    L
        !endmacro

        argA scan right to _ then argB
        argB scan right to X then continue
        continue ...

which would assemble to

        argA  _ G0001 _ R
        argA  0 argA  0 R
        argA  1 argA  1 R
        G0001 _ argB  _ L
        G0001 0 argB  0 L
        G0001 1 argB  1 L
        
        argB  _ G0002    _ R
        argB  0 argB     0 R
        argB  1 argB     1 R
        G0002 _ continue _ L
        G0002 0 continue 0 L
        G0002 1 continue 1 L

        continue...

The label 'backup' has been replaced by 'G0001' in the first macro
expansion, and by 'G0002' in the second expansion.  

There are a couple of missing features, but as I was falling asleep a
coupld of nights ago I realized that the next step would be to
implement a compiler that compiles forth to macro assembler.
        
> Next, a swap, and a rotate 3, and support for negative numbers,
> and ... why soon I'll have a Forth interpreter!

The next step seems to me to be a PostScript interpreter.  :)