@@ -5,6 +5,8 @@ Missing APIs from the types we cover (APIs have been added after this formalizat
...
@@ -5,6 +5,8 @@ Missing APIs from the types we cover (APIs have been added after this formalizat
* Structural conversion for slices. The matching operations in our model would be
* Structural conversion for slices. The matching operations in our model would be
`&mut Cell<(A, B)>` -> `&mut (Cell<A>, Cell<B>)` and
`&mut Cell<(A, B)>` -> `&mut (Cell<A>, Cell<B>)` and
`&Cell<(A, B)>` -> `&(Cell<A>, Cell<B>)` (both being NOPs).
`&Cell<(A, B)>` -> `&(Cell<A>, Cell<B>)` (both being NOPs).
* Turns out to be very hard! The way we currently associate NA-masks with locations is in conflict with this.
The invariant for the entire cell gets allocated "on" the first location of the cell, so when we do splitting the 2nd projection has no way to access it...
