Is code generation only text-based, as opposed to ...
# arrow-meta
m
Is code generation only text-based, as opposed to AST-based, currently? Or is it just the most common method in use?
r
both are common you would consider text based the quote system that allows you to generate new sources or change the tree in place but also there is just IR transformations which is codegen before bytecode
m
In my mind I was talking about source tree manipulation, the first stage, because that what I'm trying to learn at the moment. Of course, you could not know that. Sorry. So that stage is implemented using the quote system, and thus it's text-based only. I asked just in case I was missing something. For some uses, manipulating (visiting, filtering, replacing) the AST can be easier and safer. Thanks!