Is there any way to get the stacktrace when using the panic::set_hook?
I am in the unfortunate position I cannot use RUSTC_BACKTRACE env variable to understand what is causing a panic. My initial idea is to use the panic::set_hook to handle the situation and retrieve the information in a custom way. However, I just realized that using PanicInfo I only get the location of the single line causing the panic. And this is mostly useless when this comes from an unwrapping instance.
Are you looking for the backtrace crate? This is essentially what RUSTC_BACKTRACE or failure would use under the hood, and it gives you a nice programmatic way of creating and inspecting a backtrace.
Perfect, thanks!
I was not sure this was the actual crate to do that -- the 0.2 version was suspicious.
I am curious on what made you think that?
If you are asking why I was unsure about that crate, it is just because in a usual situation we have an integrated stacktrace dumping capability, and I (wrongly) expected to be based on a pretty stable API.
If you are asking why I cannot use the default RUST_BACKTRACE env variable, it is because I am doing something related to DLL injection on Windows (LD_PRELOAD, I miss you so much...), and I needed to use the MessageBoxA to get the cause of a panic.