Deprecate erlang:exit/2 and erlang:exit/3

[richcarl]

This is a followup based on #10801 and intended for OTP 30.

[github-actions]

CT Test Results

No tests were run for this PR. This is either because the build failed, or the PR is based on a branch without GH actions tests configured.

Results for commit 5095605

To speed up review, make sure that you have read Contributing to Erlang/OTP and that all checks pass.

See the TESTING and DEVELOPMENT HowTo guides for details about how to run test locally.

Artifacts

• Complete CT logs (Download Logs)
• HTML Documentation (Download HTML Docs)
• Windows Installer

// Erlang/OTP Github Action Bot

[mikpe]

Why? What problem does this solve?

[richcarl]

Why? What problem does this solve?

Apart from the small detail that exit/2 ignores signals sent by a process to itself (probably rare that this happens, but if it does, it could be quite confusing), the main thing is that exit/1 and exit/2 do very different things, one on the exception level and the other on the signal level. When reading code, it's very easy to miss that what you thought was a simple exit/1 was in fact an exit/2 or vice versa - in particular if they occur in the same module. (I've seen cases with exactly one exit/2 among a bunch of exit/1.)

So I think there is a point, for maintainability, to have a warning that steers people towards using exit_signal/2 instead. Especially if documentation and training material moves on to using the new name, calls to exit/2 will become even more obscure.

[bjorng]

There are some build errors when building eunit. Please rebase this branch on the latest master and ensure that it builds.