emacs: allow wrapped emacs to execute itself again

[binarin]

fixes #145302 #237855

emacsWithPackages wrapper script/site-start.el sanitizes EMACSLOADPATH, to make nested emacs invocations independent of the package set specified in emacsWithPackages.

But there are valid use cases when one needs to call nested emacs with the same package set. This includes built-in emacs functionality such as restart-emacs and async native compilation, and also external packages like emacs-async and esup. In all of these cases invocation-directory/invocation-name variables are being used to launch nested emacs.

With this patch these variables will be populated to point to the emacsWithPackages wrapper executable, so that executing (file-name-concat invocation-directory invocation-name) will give a fully functional emacs again.

EMACSLOADPATH sanitization was introduced by #106486, this behaviour stays unchanged. The reasoning was to be able to run different emacs executables without polluting their EMACSLOADPATH (as described here 23d4bfb). The only change is that invoking itself is again feasible (and that's what emacs actually expects).

Things done

• Built on platform(s)
  • x86_64-linux
  • aarch64-linux
  • x86_64-darwin
  • aarch64-darwin
• For non-Linux: Is sandboxing enabled in nix.conf? (See Nix manual)
  • sandbox = relaxed
  • sandbox = true
• Tested, as applicable:
  • NixOS test(s) (look inside nixos/tests)
  • and/or package tests
  • or, for functions and "core" functionality, tests in lib/tests or pkgs/test
  • made sure NixOS tests are linked to the relevant packages
• Tested compilation of all packages that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage
• Tested basic functionality of all binary files (usually in ./result/bin/)
• 25.05 Release Notes (or backporting 24.11 and 25.05 Release notes)
  • (Package updates) Added a release notes entry if the change is major or breaking
  • (Module updates) Added a release notes entry if the change is significant
  • (Module addition) Added a release notes entry if adding a new NixOS module
• Fits CONTRIBUTING.md.

---

Add a 👍 reaction to pull requests you find important.

[jian-lin]

Looks great at a first glance. I'll do a real review later.

Please target the staging branch.

[jian-lin]

One small problem: I cannot reproduce the restart-emacs issue.

My steps are as follows:

1. nix build --include 'nixpkgs=flake:github:nixos/nixpkgs?rev=55d15ad12a74eb7d4646254e13638ad0c4128776' --impure --expr 'with (import <nixpkgs> { config = { }; overlays = [ ]; }); emacs.pkgs.withPackages (ps: [ ps.magit ])'
2. export PATH=$PWD/result/bin:$PATH
3. emacs --init-directory=$(mktemp -d)
4. M-x restart-emacs RET
5. M-x magit RET

There is no error in step 5, which means the restarted Emacs can find its packages.

[jian-lin]

I took the liberty to dealt with the suffix / of invocation-directory and remove restart-emacs from commit message. If it is OK with you, I will merge this PR in 24 hours.

In addition, I re-targeted this PR to staging-next so that it can reach users sooner.

[binarin]

Yes, you are right - restart-emacs was actually not affected, only (shell-command (file-name-concat invocation-directory invocation-name)) (it's what was actually tested in #145302). I've just incorrectly assumed that restart-emacs is also broken.

I've looked a bit into why restart-emacs was working. Turns out (setenv ...) doesn't actually change OS environment, it only mutates process-environment elisp var. But restart-emacs doesn't use it, it just calls execvp in the original process env which is still pristine since after the start.