Bug report

Bug description:

That leads to flaky behaviors.

This is my PYTHONSTARTUP file:

raise ModuleNotFoundError("oops!")

ModuleNotFoundError is just an example exception that can occur in PYTHONSTARTUP scripts.

This is how the regular REPLs handle it (both the basic REPL and the new REPL):

❯ PYTHONSTARTUP=t.py ./python Python 3.15.0a1+ (heads/main:fbf0843e39e, Oct 18 2025, 11:24:43) [GCC 15.2.1 20250808 (Red Hat 15.2.1-1)] on linux Type "help", "copyright", "credits" or "license" for more information. Traceback (most recent call last): File "t.py", line 1, in <module> raise ModuleNotFoundError("oops!") ModuleNotFoundError: oops! >>> 

The traceback is shown and we proceed to the loop.

For the asyncio REPL, startup file exception always causes it to exit immediately -- we'd never want the REPL to exit because of a startup file error. On top of that, it doesn't report the exception properly, because it's racing with the interpreter shutdown at that point.

I've found 3 variants of the incorrect behavior:

1. The interactive daemon thread doesn't write the full traceback:

   ❯ PYTHONSTARTUP=t.py ./python -m asyncio asyncio REPL 3.15.0a1+ (heads/main:fbf0843e39e, Oct 18 2025, 11:24:43) [GCC 15.2.1 20250808 (Red Hat 15.2.1-1)] on linux Use "await" directly instead of "asyncio.run()". Type "help", "copyright", "credits" or "license" for more information. Exception in thread Interactive thread: exiting asyncio REPL... Traceback (most recent call last): 

2. The interactive daemon thread does flush, but after exiting is reported:

   ❯ PYTHONSTARTUP=t.py ./python -m asyncio asyncio REPL 3.15.0a1+ (heads/main:fbf0843e39e, Oct 18 2025, 11:24:43) [GCC 15.2.1 20250808 (Red Hat 15.2.1-1)] on linux Use "await" directly instead of "asyncio.run()". Type "help", "copyright", "credits" or "license" for more information. Exception in thread Interactive thread: exiting asyncio REPL... Traceback (most recent call last): File "/home/bswck/Python/cpython/Lib/threading.py", line 1075, in _bootstrap_inner self._context.run(self.run) ~~~~~~~~~~~~~~~~~^^^^^^^^^^ File "/home/bswck/Python/cpython/Lib/asyncio/__main__.py", line 104, in run exec(startup_code, console.locals) ~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "t.py", line 1, in <module> raise ModuleNotFoundError("oops!") ModuleNotFoundError: oops! 

3. The interactive daemon tries to acquire stderr at interpreter shutdown, and fails:

   ❯ PYTHONSTARTUP=t.py ./python -m asyncio asyncio REPL 3.15.0a1+ (heads/main:fbf0843e39e, Oct 18 2025, 11:24:43) [GCC 15.2.1 20250808 (Red Hat 15.2.1-1)] on linux Use "await" directly instead of "asyncio.run()". Type "help", "copyright", "credits" or "license" for more information. Exception in thread Interactive thread: exiting asyncio REPL... Traceback (most recent call last): File "/home/bswck/Python/cpython/Lib/threading.py", line 1075, in _bootstrap_inner self._context.run(self.run) ~~~~~~~~~~~~~~~~~^^^^^^^^^^ Fatal Python error: _enter_buffered_busy: could not acquire lock for <_io.BufferedWriter name='<stderr>'> at interpreter shutdown, possibly due to daemon threads Python runtime state: finalizing (tstate=0x00000000009c3818) Current thread 0x00007f6c4781cbc0 [python] (most recent call first): <no Python frame> Aborted (core dumped) 

CPython versions tested on:

CPython main branch

Operating systems tested on:

Linux

Linked PRs

• gh-140287: Handle PYTHONSTARTUP script exceptions in the asyncio REPL #140288