This is a bytecode interpreter written entirely in Windows Batch script. Right now it is incomplete and can only interpret a few Lua-style bytecode instructions.
This project was inspired by this tweet by the author of GLFW. I'm not exactly sure what the advantages of this are as opposed to just including Shell/Batch scripts in your project, but it seemed like a fun challenge.
- Some characters cannot be read in from bytecode files because of how Batch interprets them, this includes null bytes and the variable markers (% and !)
- Due to that, lots of space is wasted in the bytecode, namely the two most significant bits of every byte in the file must be
01(from most to least significant) - There is no way to natively use floating point numbers, so everything is an integer
- This thing is so ridiculously slow.
- Lua-style bytecode implementation
- Port it to POSIX Shell script (or Bash, whichever seems more reasonable)
- Write many test scripts to check for parity between Batch and Shell script implementations and to test the VM itself
- In the long term, optimize for speed and size
If you really need a cross-platform way to accompany your repository with a script which only uses interpreters built in to the OS, you should probably just use something like Batsh, a language that transpiles to both Batch and Bash. I actually plan to rewrite tinyengine in Batsh to neatly solve the portability issue this project faces, which contains about the same amount of irony as reinventing the wheel using a lathe. Not only will this interpreter likely end up taking more space than just including two Batsh-compiled scripts with your repository, but it will be significantly slower and probably less capable overall. That said, this is all for fun :)
Bitsnbites (the author of the tweet that inspired this) has their own work-in-progress implementation of this idea; however, their version is on ice for now. I guess you could say mine is the most maintained. ;)