> On Jan 16, 2024, at 10:13 AM, Johan Helsingius via cctalk
> <cctalk@classiccmp.org> wrote:
>
> On 16/01/2024 16:04, Paul Koning via cctalk wrote:
>
>> FORTH by its nature actually fits well in Harvard architectures.
>
> Indeed - but it really doesn't fit machines that aren't byte-oriented.
> I started on porting FIG-FORTH to the PDP-10 architecture but quickly
> abandoned the effort...
I suppose it's easier on byte-oriented machines but it seems doable on others.
ANSI FORTH may be helpful for this, since it explicitly distinguishes between
"character address" and "cell address" (meaning word address).
Consider also that Bowman's J-1 FORTH processor is a word-oriented machine.
His presentation slides don't even mention character words at all.
paul
