Open LDY1998 opened 3 years ago
If I remember correctly the change would be something like:
// currently
fn wrapper_for_foo(a: i32, b: bool, ...) {
foo(a, b, ...);
}
// with the change
fn wrapper_for_foo() {
foo;
}
This still lets us resolve foo
to the correct DefId
, but we don't need to worry about generics or arguments when desugaring. To be honest, I don't fully remember how I planned to deal with arguments mentioned in the specs. Anyway, the full VIR encoding will most likely not be available during your internship (as it is happening in a parallel internship), and this might be low-priority depending on how well we can desugar with the existing approach.
How is the MIR encoded process affect the decision of creating wrapper or not? Why would moving the MIR->VIR encoding into SpecCollector visit solve the wrapper problem?