I've tested the docstring of the module subprocess.
Python 2.5 is OK, but Python 3.1 fails.

~ $ python2.5 -m doctest /usr/lib/python2.5/subprocess.py
...............
----------------------------------------------------------------------
Ran 15 tests in 0.012s

OK
~ $ python3 -m doctest /usr/lib/python3.1/subprocess.py
Traceback (most recent call last):
  File "/usr/lib/python3.1/runpy.py", line 128, in _run_module_as_main
    "__main__", fname, loader, pkg_name)
  File "/usr/lib/python3.1/runpy.py", line 34, in _run_code
    exec(code, run_globals)
  File "/usr/lib/python3.1/doctest.py", line 2637, in <module>
    sys.exit(_test())
  File "/usr/lib/python3.1/doctest.py", line 2626, in _test
    failures, _ = testmod(m)
  File "/usr/lib/python3.1/doctest.py", line 1847, in testmod
    for test in finder.find(m, name, globs=globs, extraglobs=extraglobs):
  File "/usr/lib/python3.1/doctest.py", line 854, in find
    self._find(tests, obj, name, module, source_lines, globs, {})
  File "/usr/lib/python3.1/doctest.py", line 908, in _find
    globs, seen)
  File "/usr/lib/python3.1/doctest.py", line 896, in _find
    test = self._get_test(obj, name, module, globs, source_lines)
  File "/usr/lib/python3.1/doctest.py", line 980, in _get_test
    filename, lineno)
  File "/usr/lib/python3.1/doctest.py", line 590, in get_doctest
    return DocTest(self.get_examples(string, name), globs,
  File "/usr/lib/python3.1/doctest.py", line 604, in get_examples
    return [x for x in self.parse(string, name)
  File "/usr/lib/python3.1/doctest.py", line 566, in parse
    self._parse_example(m, name, lineno)
  File "/usr/lib/python3.1/doctest.py", line 636, in _parse_example
    lineno + len(source_lines))
  File "/usr/lib/python3.1/doctest.py", line 722, in _check_prefix
    (lineno+i+1, name, line))
ValueError: line 11 of the docstring for subprocess.check_output has
inconsistent leading whitespace: "'"

I fixed the docstring inconsistency (see patch) and some obvious things.

But I still have errors.


~ $ python3 -m doctest py3_failed/subprocess.py
**********************************************************************
File "py3_failed/subprocess.py", line 139, in subprocess
Failed example:
    subprocess.getstatusoutput('/bin/junk')
Expected:
    (256, 'sh: /bin/junk: not found')
Got:
    (32512, '/bin/sh: /bin/junk: No such file or directory')
**********************************************************************
File "py3_failed/subprocess.py", line 449, in subprocess.check_output
Failed example:
    check_output(["ls", "-l", "/dev/null"])
Expected:
    b'crw-rw-rw- 1 root root 1, 3 Oct 18  2007 /dev/null\n'
Got:
    b'crw-rw-rw- 1 root root 1, 3 Nov 15 17:41 /dev/null\n'
**********************************************************************
File "py3_failed/subprocess.py", line 455, in subprocess.check_output
Failed example:
    check_output(["/bin/sh", "-c",
                  "ls -l non_existent_file ; exit 0"],
                  stderr=STDOUT)
Expected:
    b'ls: non_existent_file: No such file or directory\n'
Got:
    b'ls: cannot access non_existent_file: No such file or directory\n'
**********************************************************************
File "py3_failed/subprocess.py", line 559, in subprocess.getstatusoutput
Failed example:
    subprocess.getstatusoutput('/bin/junk')
Expected:
    (256, 'sh: /bin/junk: not found')
Got:
    (32512, '/bin/sh: /bin/junk: No such file or directory')
**********************************************************************
3 items had failures:
   1 of   6 in subprocess
   2 of   2 in subprocess.check_output
   1 of   4 in subprocess.getstatusoutput
***Test Failed*** 4 failures.
~ $

It may be fixed with "# doctest: +ELLIPSIS" and "# doctest: +SKIP"
directives...
But it is probably platform-dependent.

I have tested only with Debian Lenny.

Forget the "python2.5 -m doctest /usr/lib/python2.5/subprocess.py" in
first message msg95625.
This command does not do what I guessed. And there's no doctest in the
Python2.5 "subprocess" module.


The attached patch may be enough to fix the docstring inconsistencies
with Python3.1.
It might not be mandatory to make the docstring 100% successful.
(I see no PEP which requires that).

Patch against the trunk, and against the Py3k branch.

It fixes:
 * docstring and documentation formatting
 * and examples

Patch updated to backport r51536 in trunk.
It makes "python -3" happier (do not compare None >= 0).

Committed to r76925 (trunk) and r76926 (py3k).

Thank you Georg.

I see small remaining inconsistencies in the docstring.
Either we agree with implicit "import subprocess" or "from subprocess
import *", but we should not mix both.

See additional patches:
 * issue7381_v3.diff
 * issue7381_py3k_v3.diff

OK, applied in r77086, r77087.