In the original RefCell implementation there is a cfg feature called "debug_refcell", which in case of a panic allows to see where a RefCell was first borrowed:
pub struct RefCell<T: ?Sized> {
borrow: Cell<BorrowFlag>,
// Stores the location of the earliest currently active borrow.
// This gets updated whenever we go from having zero borrows
// to having a single borrow. When a borrow occurs, this gets included
// in the generated `BorrowError/`BorrowMutError`
#[cfg(feature = "debug_refcell")]
borrowed_at: Cell<Option<&'static crate::panic::Location<'static>>>,
value: UnsafeCell<T>,
}
I read that you stripped RefCell of everything that conflicted multi-thread environment. Is it because of this that the debug feature was removed? Is it hard to reimplement it back? Pushing borrow checking into runtime leaves us with less info to investigate things. This leads projects like accountable-refcell to store the whole stack trace per RefCell. Well, that much of an overhead is probably overkill, but to have the minimal "borrowed_at" info is an essential thing to have for debug builds, don't you think?🤔
In the original RefCell implementation there is a cfg feature called "debug_refcell", which in case of a panic allows to see where a RefCell was first borrowed:
I read that you stripped RefCell of everything that conflicted multi-thread environment. Is it because of this that the debug feature was removed? Is it hard to reimplement it back? Pushing borrow checking into runtime leaves us with less info to investigate things. This leads projects like accountable-refcell to store the whole stack trace per RefCell. Well, that much of an overhead is probably overkill, but to have the minimal "borrowed_at" info is an essential thing to have for debug builds, don't you think?🤔