Correctly handle should_panic doctest attribute

[GuillaumeGomez]

Fixes #143009.

Instead of checking the exit code, we directly wrap the doctest code with catch_unwind and if a panic happened, then we return success for the test, otherwise we display the appropriate error message.

I think last one who reviewed doctest changes was @fmease so setting you as reviewer. :)

r? fmease

[rustbot]

fmease is not on the review rotation at the moment.
They may take a while to respond.

[rustbot]

This PR modifies run-make tests.

cc @jieyouxu

[GuillaumeGomez]

Ah, someone else then (sorry for the ping).

r? notriddle

[purplesyringa]

I don't like this: this causes std::process::exit(0); to be considered as a successfully passed test, and fn main() { panic!(); } is treated as an error rather than a panic.

[GuillaumeGomez]

I can't get this information from just within the same process, I think I'll create a file in case the panic happened and then check from the parent if the file exists (or any IPC).

[purplesyringa]

Can we maybe just compare the exit code with 101? It's exceedingly that unlikely that someone uses this exit code within a should_panic doctest. That feels more robust than temporary files despite the possible false positive.

[GuillaumeGomez]

Does it work on windows?

[purplesyringa]

It should, yeah. We use the same exit code regardless of the platform..

[GuillaumeGomez]

I wasn't super happy about this approach but I guess it's rare enough so that we can use it.