I've been motivate to write rs2bril like the current ts2bril. I didn't quite realize until half way through that the stalled #107 also has an implementation of a rs2bril like tool. We approached the problem in a similar way so there was some repeated work though we are targeting different extensions of Bril so this pr has some set of new contributions. The implementation structure is similar enough that https://www.cs.cornell.edu/courses/cs6120/2020fa/blog/bril-algebraic-data-types/ can be used as an overview.

#107's rs2bril targets core Bril + ADT's. ts2bril targets core Bril + Floats. This pr implements a rs2bril tool which targets core Bril + Floats + Memory. I've added a full list of limitations that I'm aware of to the README, I will highlight a few:

• Currently requires all variable declarations to be explicitly typed.
• println! is expected to be valid rust unlike in Algebraic types extension #107, however the format string is ignored during compilation and you can only pass variables as arguments.
• It always assumes valid Rust code(you can run all of your code as Rust first!), except in the case that you want to pass arguments to main. In this case, just declare function arguments to main like any other function(which makes it not valid Rust).
• Memory is implemented as Rust Arrays/Slices. This means that they are statically sized with integer literals and must be initialized on creation. I've also specialized drop to compile to free. Programmers need to manually manage their memory(by dropping arrays). This could be a use case for https://www.cs.cornell.edu/courses/cs6120/2020fa/blog/asmm/ .

It can provide source positions via the -p flag though many are probably off/missing.

example.rs has some example syntax that is supported. I've added a set of tests which are equivalent to the ts2bril tests plus more complicated benchmarks like ackermann.rs, riemann.rs and cholesky.rs(which is probably the best representation of what rs2bril can do).