Post by @dirtycommo@anticapitalist.party

Dandelion BASIC devlog

I can see the issue I am having is because I am trying to follow the source code of the original 8080 Tiny BASIC so closely.

I do not necessarily need to copy every single detail, but, the process has been rewarding. In the end I was able to learn how to do macros.

Given I have my tool chain set up nicely, perhaps I could experiment with my own execution engine.

The Tiny BASIC should be based around what is useful to the project.

🧵

3 Replies

3 Boosts

The interpretive language, IL, method has the benefit that you can rewrite the IL without having to resort to getting into complex assembly.

I want to have 2D arrays, and FOR-NEXT.

This is supported by the TBX IL program.

I will try and experiment with different opcode encoding schemes. I may just keep the original two bit scheme, MLCALL, IJMP, ICALL, and TST.

What I need is some way to point the interpreter at the heading with the machine language code that names up the IL routine.

🧵

Perhaps having a token for every ML routine, a kind of massive expansion of the two bit opcode scheme, would be wise.

That way, the execution engine just keeps a jump table, and then dispatches control to the matching ML routine.

Sounds a little bit more like Li Chen Wang's version, but this time we keep our IL.

When the CPU sees the IL opcode token, it jumps to the relevant corresponding routine like the IL method would normally expect.

Okay, I think that is the solution.

@dirtycommo a 2d array is just a linear array. The trick is in the API: use two indexes to compute the linear offset.