Setup / Include File / Assembler / Compiler etc

[adrianpbrown]

So lets get some opinions going.

1. Should all the code be written to work with the same assembler?
2. Should we agree on a standard header file for defining things such as IO_VMPR so all teh code is consistant
3. Do we want a specific structure to where things are, so folders for different things
4. Should it just be snippets, e.g. i have a maths.asm that has various maths related functions, do i upload that or do we just have single files for 16bit MUL etc

[AntonyDTA]

1. I don't know.
2. It's not necessary for me, I prefer to use a number and if necessary I'll put its meaning in a note.
3. I don't know exactly what you mean, but it will definitely need to be organized in some folders according to content.
4. Both should be possible. In the Math/Mul/ folder there can only be different variations of Mulu, Muls, 8_8_16, 8_16_24, 16_16_32, etc. And in the Math/Lib/ folder there can be files with multiple functions that have common parts of the program, or tables.

[spectecjr]

I'd keep actual routines and snippets separate from ports/hardware headers. They should probably go in repos defined by category (e.g. "math-routines", "text-routines", "keyboard-routines"), so that people can consume specific versions as they might change over time.

There's two kinds of usage - macros (so INCLUDE "" the macro, and then use it elsewhere), and full source file inclusion. (I imagine most will come in both flavors so they can be directly inlined; so the source file just includes the macro and uses it once with a named entrypoint).

Gets trickier with data, so we need readmes for each routine.

Specific assembler... god.. that's a hard one. I'm writing one which abandons a lot of the COMET era syntax. I think what we might end up with is some kind of machine translation between different assemblers, and try to specify a specific core.

For now we might just want to have folders which clearly identify which assembler people are targeting. I imagine most people will use Pyz80 until there's something better. We can just roll with the punches for that.

IMO the goal is to shift development to modern computers and output SAM-compatible stuff, so having a ton of files should probably be the norm - the penalty associated with it is minimal - and that way you end up with logically grouped routines in a single file, but most of the time, one routine per file.

Worst case people can always download it, and then cut & paste into their project.

Just my 2p.

[davidimpress]

Also wondering about adding (with permission) routines from other None-SAM based projects as well.