[simias]

I don't think Rust can notify on a failing destructor other than panic!ing. AFAIK the best you can do if you want to handle errors on close is to call `flush()` (which does return errors) before dropping the object. Of course that nullifies the benefits from RAII.

I don't know if there's an elegant way to solve this. If Rust had exception you could use that but then again in C++ it's often explicitly discouraged to throw in destructors because you could end up in a bad situation if you throw an exception while propagating an other one. How does Python's "with" handle that?

[masklinn]

> I don't think Rust can notify on a failing destructor other than panic!ing.

Much as in C++, this is not really allowed: drop runs during panic unwinding, a panic during a panic will hard-abort the entire program.

> I don't know if there's an elegant way to solve this.

I don't really think there is. Maybe opt-in linear types could be added. That would be at the cost of convenience (the compiler would require explicitly closing every file and handling the result, you could not just forget about it and expect it to be closed) but it would fix the issue and would slightly reduce the holding lifetime of resources.

Furthermore, for convenience we could imagine a wrapper pointer converting a linear structure into an affine one.

> How does Python's "with" handle that?

You'll get the exception from `__exit__` chaining to whatever exception triggered it (if any). Exceptions are the normal error-handling mechanism of Python so it's not out of place.

[simias]

>Much as in C++, this is not really allowed: drop runs during panic unwinding, a panic during a panic will hard-abort the entire program.

Right, I didn't really consider that a "drawback" because I'm in the camp that considers that panic! shouldn't unwind but actually abort the process here and there anyway. But you're right that if you rely on the default unwinding behavior panic!ing in destructors is a very bad idea.

[cesarb]

> But you're right that if you rely on the default unwinding behavior

You do rely on the default unwinding behavior anyway at least for `cargo test`: the test framework depends on being able to catch the unwinds from `assert_eq!` and similar.

[dllthomas]

> I don't know if there's an elegant way to solve this.

It could take a callback. Then for any given file handle, if you don't care that the write failed you can ignore it; if you care but can't sensibly respond, you can panic; if you can sensibly respond you can do it inline or schedule work to be done somewhere with a longer lifetime than the file handle.