Here, we all know that the fasm2 CALM language is "assembly written interpreter/compiler" friendly for obvious reasons.

How much "final" is CALM language in order to write assemblers with support of classic object formats (ELF/PE+/etc)?

For CPUs, I guess RISC-V (with its peculiarities) could be implemented without nasty hacks using this macro language.

On the GPU side, I could bet an AMD RDNA assembler would be free from nasty hacks too.

Should CALM get an RFC?

What specific features or support does CALM currently lack that you feel are critical for handling ELF/PE+?

What would be the goals or criteria for a CALM RFC?

For me, CALM doesn't make sense outside of its supporting macro language. The macro language language makes easy tasks which are very complex for CALM and visa-versa.

If you exclude the classic macro language, sure some things will require a bit more work. But if I recall properly, you only need CALM.

And RFC literaly means to force stability on the long run.

Anything I would consider cruft in fasmg is usually the result of backward compatibility - it so backward compatible that I'd need to search for few cases that are problematic. It's must one of Tomasz's top priorities - if not the top. Getting more stable than that would require no further development, imho.

We could test it out - there are plenty of old codes here on the board.

On the long run, elite projects like this one usually factors in only pertinent usage for their task, removing what were a bad ideas or 'too much', and usually it kind of gets stable in time by itself, if people in charge are reasonable. Backward compatibility is usually just 'glue' to what in the end is pertinent and well balanced.

That's why I am looking at CALM and not the classic macro language. I dare to think that what I am looking for may actually be CALM alone.

CALM is minimal, CALM is 'assembly written interpreter/compiler' proof... and it does a good enough job for file formats and various ISAs.

That said, it has been a while I have not touched fasm. I am currently more into RISC-V assembly (no pseudo-ops) and I use a basic C preprocessor for this code (not from those c++ abominations which are gcc/clang ofc).

But in my head (even if now far away in time) I still have some fuzzy memories of consistency of CALM for writting assembler AND be a powerful macro language.

I forget who said it, "The hardest thing about being a painter is putting the paint down."
- on the canvas
- away from the canvas
- and finally

The Three Surrenders

First surrender:
brush meets canvas
trembling with possibility
and terror

Second surrender:
stepping away before
perfection destroys
what honesty created

Final surrender:
laying down the brush
letting the work
speak for itself
while we
at last
rest

... coding can be the same.

....

"And a piece of software is finished when there is nothing else to _remove_."

That said, if I am not too mistaken, x86 fasm2 assembler is being rewritten using CALM?