The Sinclair QL Forum

Hi,

Which is the best Forth to use?

NormanDunbar wrote:Funnily enough, I find that when I get a new CPU to play with, I immediately start looking for Assembly Language stuff! I have to admit, either I don't look and play hard enough with ARM assembly, or it's just too weird for my ageing brain! I suspect the latter.

Most later ISAs were developed for compilers and not for humans. The pinnacle of this trend was probably the Itanium, which was more or less impossible to write Assembler code for. Instructions always came in chunks of 3 and had certain rules to them: https://devblogs.microsoft.com/oldnewth ... 0/?p=90811

Derek_Stewart wrote:Hi,

Which is the best Forth to use?

The question is : on what machine and for what ? Classic FORTH machines were 16bits on 64Kb (or less) memory, and could be (and were) metacompiled for almost all processors, following basically very similar models (norms : 79, fig, 83...). What this programming environment was typically good for was driving complex automation processes (first with minicomputer, after with microprocessors, and nowadays even with microcontrolers).

https://arduino-forth.com/ (I personnaly know Marc Petermann who did this)

Here you have an alternate development tool for arduino, using FORTH instead of C and performing rather differently : instead of developping with incessant C compilations and uploads between a host system and the arduino target, the FORTH kernel is permanently resident and capable to compile new words interactively through the serial line and a terminal. So you may test and debug interactively and progressively your applications word by word on the target itself.

Nowadays, there are indeed more powerful FORTH systems, like gFORTH for linux, etc. Todo : try them on my Raspberries ; and try other FORTHs (brouhahouba) for the QL.

polka wrote:My first "personnal" computer was a SCMP elektor kit that you could only program with machine code through a hex keyboard,.

Off and on over the last 20yrs I have written about 24K of SCMP code to control analogue synth (ETI/Maplin 5600) and improved Elbug. Didn't code it in hex though, I wrote an assembler and disassembler for INS8060 in QL SuperBASIC. I was toying with idea of writing a Forth for it but ran out of steam (my clone of Elektor system has 16K RAM, 32K paged EPROM, INS8154 for PIO). Never found any info on Forth for INS8060 (very expensive chip now).

NormanDunbar wrote:

polka wrote:... I am not interested in assembler programming.

Sacrilege!

Funnily enough, I find that when I get a new CPU to play with, I immediately start looking for Assembly Language stuff!.

Ever looked at TMS9900 CPUs? (TI-99/4A computer), they are odd*. I almost bought a Cortex computer** (TMS9995) in mid 80's but thankfully got distracted by QL. Someone did an emulator of Cortex and I tried a bit of assembly coding: no stack/registers in ram/context switching - but quick CPU. I did a comparison using Basic and the Cortex seemed a bit quicker machine than standard QL.

* even the address and data lines on CPU are reversed (A0/D0 are most significant)

** https://www.stuartconner.me.uk/powertra ... cortex.htm

mk79 wrote:Most later ISAs were developed for compilers and not for humans. The pinnacle of this trend was probably the Itanium, which was more or less impossible to write Assembler code for....

Luckily, or nicely done, I've never had to write assembly for anything resembling an X86/Pentium or Itanium. Sounds like a lucky escape though.

Cheers,
Norm.

I remember having to write 370 Assembler for the tills at Harrods - it was originally a macro, but someone expanded it out to the source code and then started editing it directly - so it was a mare to maintain.

I probably should have sought therapy after that job!