It's an honor to hear from you, BDFL.

I'm just a Python user, not a developer, so the time it would take me
to check out, compile, test, etc. would dwarf the change itself
(removing two lines from file_dealloc).

There is another solution, following Java: file_dealloc could raise an
exception. Then the error would appear in normal contexts, but
(assuming the GC ignores exceptions in finalization) would be ignored
in GC.

There is also a simpler, less invasive change you could consider:
disabling GC in the child inside subprocess.Popen.

On 10/26/07, Guido van Rossum <report@bugs.python.org> wrote:
>
> Guido van Rossum added the comment:
>
> I don't think we can prevent GC from occurring between fork and exec --
> it's legal to just call os.fork() and execute Python code in the
> subprocess forever.  I think the right solution might be to ignore
> errors in file_close().  Can you try to whip up a patch for that and
> test it?
>
> ----------
> nosy: +gvanrossum
>
> __________________________________
> Tracker <report@bugs.python.org>
> <http://bugs.python.org/issue1336>
> __________________________________
>