Cool. Why, though? I mean, it's not like this project has any practical value anyway, but wouldn't it be actually easier and more efficient to actually implement a C compiler in VimScript? Or is the vimscript ELVM backend the actual end-goal here, and C-compiler at question is basically just a proof that it works?
Thanks, at first I thought someone had written it in vim script, then I inspected the README in detail and learned it's only transpiled from C to vim script, which is not particularly exciting or impressive compared to a human wielding such a degree of vim script fu.
I wonder how much faster a manually written C compiler in Vim script would be.
The current code is essentially assembly instructions running as Vim script, this gotta be incredibly slow, probably not much faster than an x86 emulator written in Vim script:
Funny because neovim supports LuaJIT for it's lua init files by default. I wonder if there would be a way to compile vimscript to lua and have it everything inside the lua IR. Probably very hard, given the DSL nature of vimscript
This was tried by ZyX-I when the Neovim project was started. AFAIK they managed to create an alternative parser for Vimscript, but the transpiler was never finished (or productionized?), I think due to incompatibilities that are hard to avoid. Some relevant issues:
