[NgoQuocViet2001]

When the command started by dotenv run is killed by SIGPIPE, the CLI never exits. Native tools get SIGPIPE when the reader of a pipe stops early, for example dotenv run -- cat app.log | head or dotenv run -- grep ERROR app.log | head on a large file. The output is printed, then the pipeline hangs until Ctrl-C or a CI timeout. The same command without dotenv exits with 141.

The exit handler re-raises the child's signal on itself and keeps the event loop alive with a setInterval until that signal arrives. Node ignores SIGPIPE, so it never arrives and the process stays alive forever. This replaces the interval with a 1s setTimeout that exits with the shell's 128 + signal number status. Signals Node does not ignore still terminate the process by signal as before. SIGXFSZ had the same hang and now exits 153.

Ran:

• npm test on Windows: 1412 pass, 1 skip (POSIX signal tests skip on win32)
• full tap suite on Linux: 1418 pass, 1 skip
• new test in tests/test-cli-signals.js: fails on master with ETIMEDOUT, passes with the fix

Checked:

• cat/grep on a 1M-line file piped to head -n 1: PIPESTATUS 124 0 (timeout) on master, 141 0 with the fix
• sh -c 'kill -TERM $$' still exits 143, sh -c 'kill -KILL $$' 137, sh -c 'exit 7' 7

[motdotla]

nice catch