cranelift VS pest

(And then I'd turn them into normal BBs internally, since cranelift found out they're not that great for optimizing: https://github.com/bytecodealliance/cranelift/issues/796.)

I was joking but thats almost what they are doing. Instead of using LLVM they're using cranelift for JITting

Replace it with what, and why? There are codegen backends using Cranelift and GCC, but they are not feature complete yet.

Thanks for taking the time to provide a detailed response. Need to think on this. I had actually started a regular Clojure parser using LLVM as a hobby, but then my friend said it was better to implement this in a safe-memory and. save-concurrency language like Rust and leverage cranelift for code generation. So, now I am learning rust, lol. (Btw, Rust has a persistent data structures lib too: https://github.com/orium/rpds )

I have been on the lookout for something that code help me with Jinja and it seems like a plausible candidate. Extra bonus points if I could actually embed the Python in the same grammar but at this point even having the code blocks segregated from the literal parts is a good start

https://pest.rs/?g=N4Ig5gTghgtjURALhAMwJYBsCmACAvLsLgMoDyAkr...

I'll throw in a plug for https://pest.rs/ a PEG-based parser-generator library in Rust. Delightful to work with and removes so much of the boilerplate involved in a parser.

Would definitely be interesting, but from a cursory look at the repository, it doesn't look like squeezing the last percentage points of performance has been a priority yet.

Things that stand out:

- The `awk` implementation uses the Pest parser generator (https://pest.rs/), which is known to not generate the fastest possible parsers, but is great for getting up and running.

- They are using the `clap` crate for argument parsing, which is also known to not be the fastest, but again is very user friendly (for example, it does Unicode linebreaks in the output of `--help`). It's marginal, but for a tiny utility being invoked many times from a shell script, this can add up.

It's very probably "fast enough", and it makes sense to prioritize like this at this point, but people shouldn't use this expecting a performance improvement right now.

Nice work. Out of curiosity, did you consider using pest (https://pest.rs/) to help build your parser? Or is it too much for what you are doing?

And some related parser tools: - https://github.com/kevinmehall/rust-peg - https://github.com/pest-parser/pest - https://github.com/lalrpop/lalrpop

I had recently completed the first year of my Computer Science class at school and will begin my second year soon. My schools' class forces the use of Java programming language, and I absolutely hated it. So, over the course of a little less than a month, I wrote my own programming language, in Rust (objectively best programming language), using pest, to be as similar to Rust as possible, but compiling to Java.

I second pest.rs. Using it is fairly intuitive and there's also a live playground on their website which is great for quickly developing and testing your AST (abstract syntax tree) parser for whatever language you're implementing.