Acornsoft ISO Pascal object-code format?

[nicolagiacobbe]

Hello everybody.
Sorry for asking help but to date I have been unable to find what seemed a simple piece of information. I am playing with Acornsoft ISO pascal and it is working great (with some quirks) but it is a black box to me. I would like to use the pascal interpreter alone as a virtual machine on ROM and make a few tests to see how efficient could be in term of code space.
Anyone pointing me to some resource to understand the object format and the underlying VM?

[SKS1]

Hello everybody.
Sorry for asking help but to date I have been unable to find what seemed a simple piece of information. I am playing with Acornsoft ISO pascal and it is working great (with some quirks) but it is a black box to me. I would like to use the pascal interpreter alone as a virtual machine on ROM and make a few tests to see how efficient could be in term of code space.
Anyone pointing me to some resource to understand the object format and the underlying VM?

Have a look at its compiler? Or rather, one that's as close as you can get given the actual source is lost in the Acorn Drawing Office.

https://gtoal.com/acorn/arm/TutuPascal/AComp/

I was adapting the 6502 one for ARM in late 1985/early 1986, having done a 65816 version earlier in 1985 for the Communicator, as well as bug fixing the 6502 one that was written by Ben and Li (hence BL code). I'd adapted it fairly carefully for wider pointers etc.: see comments containing "Tutu - was" ...

[nicolagiacobbe]

Thanks SKS1, now that you pointed me to it I think is the best way, just hoping for something more specific but this is already great.

[SKS1]

Thanks SKS1, now that you pointed me to it I think is the best way, just hoping for something more specific but this is already great.

I think the 6502 optimiser is lost as well; that was needed to be able to fit the BL coded compiler into a single ROM, otherwise it's about 20KB. It scanned the compiler output for commonly-found sequences and squeezed them into previously unused BL codes. I also think the disc-based compiler was slightly more functional.

I don't recall seeing a disassembly of the 6502 BL code interpreter here?

[nicolagiacobbe]

neither do I, it would be nice. It could be solved but I fear it could become another rabbit pit...

[dominicbeesley]

I'd be interested in finding out more about the 816 version. Is that lost?

[SKS1]

Thanks SKS1, now that you pointed me to it I think is the best way, just hoping for something more specific but this is already great.

I think the 6502 optimiser is lost as well; that was needed to be able to fit the BL coded compiler into a single ROM, otherwise it's about 20KB. It scanned the compiler output for commonly-found sequences and squeezed them into previously unused BL codes. I also think the disc-based compiler was slightly more functional.

Ah, there seems to be a copy of the optimiser on Graham's site, as well as the BL code disassembler

https://gtoal.com/acorn/arm/TutuPascal/BBCPascal/

[SKS1]

I'd be interested in finding out more about the 816 version. Is that lost?

As far as I am aware, yes. Worked OK, interpreter stayed in 16-bit mode all the time, using 24-bit pointers everywhere for all Pascal objects.

[Coeus]

Ah, there seems to be a copy of the optimiser on Graham's site, as well as the BL code disassembler

Are these files in BL-code? They don't look like source.

[SKS1]

Ah, there seems to be a copy of the optimiser on Graham's site, as well as the BL code disassembler

Are these files in BL-code? They don't look like source.

They're Beeb executables, can't remember whether they are run under the BL interpreter.