close() seems to have limited effect

[smontanaro]

BPO	4604
Nosy	@smontanaro, @amauryfa, @pitrou, @tiran
Files	iobug.diff io.patch io_fixes.patch: io module fixes with test

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = <Date 2009-01-09.20:08:25.385>
created_at = <Date 2008-12-09.00:11:20.450>
labels = ['extension-modules', 'type-bug', 'library', 'release-blocker']
title = 'close() seems to have limited effect'
updated_at = <Date 2009-01-09.20:08:25.384>
user = 'https://github.com/smontanaro'

bugs.python.org fields:

activity = <Date 2009-01-09.20:08:25.384>
actor = 'pitrou'
assignee = 'none'
closed = True
closed_date = <Date 2009-01-09.20:08:25.385>
closer = 'pitrou'
components = ['Extension Modules', 'Library (Lib)']
creation = <Date 2008-12-09.00:11:20.450>
creator = 'skip.montanaro'
dependencies = []
files = ['12306', '12309', '12450']
hgrepos = []
issue_num = 4604
keywords = ['patch']
message_count = 9.0
messages = ['77376', '77440', '77446', '77450', '77455', '77475', '77476', '78295', '79498']
nosy_count = 5.0
nosy_names = ['skip.montanaro', 'hdima', 'amaury.forgeotdarc', 'pitrou', 'christian.heimes']
pr_nums = []
priority = 'release blocker'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue4604'
versions = ['Python 3.0']

[smontanaro]

Seems like file.close() in 3.0 isn't much of a barrier to further reading:

% python3.0
Python 3.0rc3+ (py3k:67338M, Nov 22 2008, 06:47:23) 
[GCC 4.0.1 (Apple Inc. build 5465)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> fp = open("/etc/services")
>>> ct = fp.read(1048)
>>> print (ct[-80:], end='') ; fp.close() ; ct = fp.read(17) ; print (ct)
compressnet       2/udp     # Management Utility
compressnet       2/tcp     # Management Utility

The second read() should raise an exception. Same code, 2.6:

% python2.6
Python 2.6+ (release26-maint:66821:66833, Oct 30 2008, 22:16:1) 
[GCC 4.0.1 (Apple Inc. build 5465)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> from __future__ import print_function
>>> fp = open("/etc/services")
>>> ct = fp.read(1048)
>>> print (ct[-80:], end='') ; fp.close() ; ct = fp.read(17) ; print (ct)
compressnet       2/udp     # Management Utility
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ValueError: I/O operation on closed file
compressnet       2/tcp     # M>>> 

Culled this from a posting to comp.lang.python.

[pitrou]

Indeed, it seems to happen if you first call read() before calling close().

[smontanaro]

Here's a test case (diff against Lib/test/test_io.py). This fails for
me on OS X 10.5.5 with the tip of the py3k branch.

[smontanaro]

fixed patch. I'm still a bit clumsy with the assertRaises stuff.

[smontanaro]

Here's a minimal patch to BufferedReader.read() which causes the test
to pass. I will leave it for smarter people to decided whether or not
all the other read() methods need the same test.

[amauryfa]

• The patch corrects a specific case, but not others: for example if the
  file is opened in "r" (text) mode.
  There are also other methods to test: peek(), read1()...

• It should use
  self._checkClosed()
  which already raises the same exception with the same message.

[smontanaro]

Amaury> - It should use
Amaury> self._checkClosed()
Amaury> which already raises the same exception with the same message.

I think some other places will need this change then.

Note that I don't know the io code at all. I was just trying to provide a
simple test case and a simple fix. Feel free to generalize it as necessary.

Skip

[hdima]

Attached more generic version of the patch.

[pitrou]

Expanded the test a bit and committed the patch in r68454. Thanks!