The attached script behaves differently on Python 2.7.2 and Python 3.2.3.

With Python 2.7 the script runs for ca. 30 seconds and then I get back my prompt.

With Python 3.2 the script runs in the background, I get back my prompt immediately and can type shell commands.

Commenting out the call to uname() changes the behaviour of the script on Python 3.2 so that it behaves like on Python 2.7.

(This happens on both Max OS X 10.7 and Linux.)

The problem is that os.wait() is returning when the wrong process exits.

You can fix this by specifying the pid you are waiting for by doing "os.waitpid(pid, 0)" instead of "os.wait()".

Arguably os.popen() and subprocess.communicate() etc should always reap the pid of the process they started.

Actually, if you replace

	print(os.popen("uname").read())

with

        f = os.popen("uname")
        print(f.read())
        f.close()

or

        with os.popen("uname") as f:
            print(f.read())

then things should work.  This is because f.close() waits for the pid.

In Python 2.x, when the file object returned by popen() is garbage collected the process is automatically waited on, collecting the pid of the process.

In Python 3.x a wrapper object is used whose close method wait on the pid.  This close method is *not* invoked when the wrapper is garbage collected.  However, the set of dead pids cannot accumulate since dead pids get collected whenever os.popen()/subprocess.Popen() is used.

The old behaviour is only an advantage in a refcounted implementation of Python like CPython.  I am not sure that "fixing" the current behaviour is necessary.

Simply adding to the wrapper class a __del__() method which invokes close() will not work.  (One could instead use a weakref callback, but that is non-trivial.)

So is this simply a documentation issue, or can we close the bug as "won't fix"?

A program which depends on the old behaviour would be broken on a non-refcounted implementation of Python, so I would be inclined to say "won't fix".

However, I think the following patch would restore the old behaviour

diff -r a970054a93fb Lib/os.py
--- a/Lib/os.py	Mon Jul 16 18:30:03 2012 +0100
+++ b/Lib/os.py	Mon Jul 23 19:12:38 2012 +0100
@@ -1004,6 +1004,13 @@
 # Helper for popen() -- a proxy for a file whose close waits for the process
 class _wrap_close:
     def __init__(self, stream, proc):
+        # proc should be waited on when stream is garbage collected
+        import weakref
+        def callback(wr):
+            proc._weakref = None           # break ref cycle
+            proc.wait()
+        proc._weakref = weakref.ref(stream, callback)
+        proc.stdin = proc.stdout = None    # proc must not keep stream alive
         self._stream = stream
         self._proc = proc
     def close(self):

Given that you both nearly agreed to "wont fix" 8 years ago and 2.7 is now EOL, I'm going to go ahead and close the issue that way :)