PHP中类析构函数不被调用的脚本终止场景有哪些？

Great question—ensuring your cleanup code runs reliably is super important, especially when dealing with resources like file handles, database connections, or locks. Let’s break down the key scenarios where user-defined destructors won’t execute when a script terminates, along with contrasting cases where they do work as expected:

Destructors Will NOT Be Called #

These are cases where the runtime or operating system cuts off execution immediately, leaving no time for cleanup routines:

• Fatal interpreter errors: Things like segmentation faults (SIGSEGV), invalid memory access, or using a broken extension that crashes the VM. The interpreter can’t safely proceed to run destructors in this state.
• Forced termination via SIGKILL: The OS sends this signal to immediately kill the process—no signal handling or cleanup code gets a chance to run. This is different from SIGTERM, which allows graceful shutdown.
• OS-initiated out-of-memory (OOM) kills: If your process consumes too much memory and the kernel terminates it to save the system, there’s no opportunity for the runtime to execute destructors.
• Low-level immediate exit calls: Using system-level functions like _exit() (instead of the standard language exit()), which bypass the runtime’s shutdown sequence entirely.
• Hard system failures: Power loss, kernel panics, or hardware crashes will stop the process instantly, with no cleanup possible.

Destructors Will Be Called #

These are scenarios where the runtime gets a chance to properly shut down and run cleanup code:

• Unhandled exceptions: Even if an uncaught exception terminates the script, most languages will still trigger destructors for objects that were in scope before the exception occurred.
• Graceful signals: Signals like SIGTERM (sent by kill command by default) or SIGINT (Ctrl+C) are typically caught by the runtime, which then proceeds to run shutdown routines including destructors.
• Normal script completion: When the script finishes executing all code, destructors are called as objects go out of scope or the runtime cleans up its environment.
• Standard exit()/die() calls: Most high-level language exit functions trigger the runtime’s shutdown process, which includes running destructors for existing objects.

Quick Note on Critical Cleanup #

If you have cleanup tasks that absolutely must run (e.g., releasing a shared lock), keep in mind that some scenarios (like SIGKILL or OOM kills) are unavoidable. For these cases, consider:

• Using signal handlers to catch catchable signals and trigger cleanup before exiting.
• Adding checks in your next script run to detect leftover resources and clean them up.
• Using process supervision tools to monitor your script and handle failures gracefully.

内容的提问来源于stack exchange，提问作者Zhigalin