Jesse A. Tov
I wonder if the `sumFold` example would be clearer if it were written using `foldr` instead of with `foldl`. In particular, because `sumManual` is written with the same kind of...
We check that we can call a `ClosureOnce` once successfully, but calling it a second time is an error. Normally if this error occurs, the program exits (code 2), because...
This is intended to fix a soundness bug, the first item in #34, namely: > It erases all auto trait information attached to closure captures (read: `Send` and `Sync`) The...
Make some types Send + Sync: - middle::Cif - middle::Type - high::Type I believe this is safe because none of these types provides any kind of interior mutability. The wrappers...
Changes the argument of `ClosureMutN::code_ptr` to be `&mut self` instead of `&self`. This is an attempt to resolve the second check box of #34: > It erases mutability required to...
Right now it’s `repr(C)`, and I’m not sure whether that’s strictly correct.
Currently we recognize only `FFI_BAD_ABI` and `FFI_BAD_TYPEDEF`, which we translate to `Error::Typedef` and `Error::Abi`. As of now, `FFI_BAD_ARGTYPE` gets translated to `Error::Abi` (which maybe isn’t strictly incorrect?). But it would...
Looks like you’ve done an impressive job here. There are two things limiting this to a ✓: 1) Your testing appears perfunctory—this is not what testing well looks like. 2)...
Thanks for submitting your spec. It’s a good start, but it’s missing a lot of detail. First, in order to describe the API, you should be describing it in Rust,...