]> granicus.if.org Git - python/log
python
10 years ago(Merge 3.4) Issue #20957: test_smtpnet now uses support.transient_internet() to
Victor Stinner [Wed, 3 Sep 2014 22:36:09 +0000 (00:36 +0200)]
(Merge 3.4) Issue #20957: test_smtpnet now uses support.transient_internet() to
call check_ssl_verifiy(), so only test_connect_using_sslcontext_verified() is
skipped if smtp.gmail.com cannot be joined, not the whole file.

10 years agoIssue #20957: test_smtpnet now uses support.transient_internet() to call
Victor Stinner [Wed, 3 Sep 2014 22:35:43 +0000 (00:35 +0200)]
Issue #20957: test_smtpnet now uses support.transient_internet() to call
check_ssl_verifiy(), so only test_connect_using_sslcontext_verified() is
skipped if smtp.gmail.com cannot be joined, not the whole file.

10 years ago(Merge 3.4) Issue #22332: test_multiprocessing_main_handling is now skipped if
Victor Stinner [Wed, 3 Sep 2014 21:48:37 +0000 (23:48 +0200)]
(Merge 3.4) Issue #22332: test_multiprocessing_main_handling is now skipped if
sem_open implementation is broken (ex: skipped on FreeBSD 6.4).

10 years agoIssue #22332: test_multiprocessing_main_handling is now skipped if sem_open
Victor Stinner [Wed, 3 Sep 2014 21:48:08 +0000 (23:48 +0200)]
Issue #22332: test_multiprocessing_main_handling is now skipped if sem_open
implementation is broken (ex: skipped on FreeBSD 6.4).

10 years ago(Merge 3.4) Issue #22331: Skip test_interrupted_write_text() on FreeBSD older
Victor Stinner [Wed, 3 Sep 2014 21:33:43 +0000 (23:33 +0200)]
(Merge 3.4) Issue #22331: Skip test_interrupted_write_text() on FreeBSD older
than 8.0

10 years agoIssue #22331: Skip test_interrupted_write_text() on FreeBSD older than 8.0
Victor Stinner [Wed, 3 Sep 2014 21:32:28 +0000 (23:32 +0200)]
Issue #22331: Skip test_interrupted_write_text() on FreeBSD older than 8.0

10 years agoIssue #21933: Merge with 3.4
Terry Jan Reedy [Wed, 3 Sep 2014 20:18:34 +0000 (16:18 -0400)]
Issue #21933: Merge with 3.4

10 years agoIssue #21933: Users can now change the font size for example code.
Terry Jan Reedy [Wed, 3 Sep 2014 20:17:41 +0000 (16:17 -0400)]
Issue #21933: Users can now change the font size for example code.
Original patch by Lita Cho.

10 years agoIssue #22043: Fix pymonotonic(), use tv_usec=-1 as a marker to skip
Victor Stinner [Wed, 3 Sep 2014 07:43:48 +0000 (09:43 +0200)]
Issue #22043: Fix pymonotonic(), use tv_usec=-1 as a marker to skip
the monotonic test

10 years agoIssue #22043: time.monotonic() is now always available
Victor Stinner [Tue, 2 Sep 2014 21:18:25 +0000 (23:18 +0200)]
Issue #22043: time.monotonic() is now always available

threading.Lock.acquire(), threading.RLock.acquire() and socket operations now
use a monotonic clock, instead of the system clock, when a timeout is used.

10 years agoIssue #22043: Fix _PyTime_gettimeofday() if HAVE_GETTIMEOFDAY
Victor Stinner [Tue, 2 Sep 2014 21:01:40 +0000 (23:01 +0200)]
Issue #22043: Fix _PyTime_gettimeofday() if HAVE_GETTIMEOFDAY

Ensure also that the tv_usec field is consistent: in range [0; 999999].

10 years agoCloses #22258: Fix the the internal function set_inheritable() on Illumos.
Victor Stinner [Tue, 2 Sep 2014 09:41:04 +0000 (11:41 +0200)]
Closes #22258: Fix the the internal function set_inheritable() on Illumos.
This platform exposes the function ioctl(FIOCLEX), but calling it fails with
errno is ENOTTY: "Inappropriate ioctl for device". set_inheritable() now falls
back to the slower fcntl() (F_GETFD and then F_SETFD).

10 years agoRemove unused imports.
Berker Peksag [Mon, 1 Sep 2014 09:33:12 +0000 (12:33 +0300)]
Remove unused imports.

10 years agoIssue #19447: Suppress output of py_compile.compile().
Berker Peksag [Mon, 1 Sep 2014 09:29:53 +0000 (12:29 +0300)]
Issue #19447: Suppress output of py_compile.compile().

10 years agoIssue #22320: Fix broken link in the General Python FAQ.
Ned Deily [Mon, 1 Sep 2014 07:06:18 +0000 (00:06 -0700)]
Issue #22320: Fix broken link in the General Python FAQ.
Original patch by Josh Lynn.

10 years ago#22315: Use advertised API for OSError
Jason R. Coombs [Sun, 31 Aug 2014 21:42:20 +0000 (17:42 -0400)]
#22315: Use advertised API for OSError

10 years ago#22315: Use an existent directory for 'src' to trigger appropriate behavior.
Jason R. Coombs [Sun, 31 Aug 2014 21:51:22 +0000 (17:51 -0400)]
#22315: Use an existent directory for 'src' to trigger appropriate behavior.

10 years ago#22315: Provide an actual directory during test invocation.
Jason R. Coombs [Sun, 31 Aug 2014 21:37:35 +0000 (17:37 -0400)]
#22315: Provide an actual directory during test invocation.

10 years ago#22315: Use technique outlined in test_file_util
Jason R. Coombs [Sun, 31 Aug 2014 21:31:32 +0000 (17:31 -0400)]
#22315: Use technique outlined in test_file_util

10 years ago#22315: Add test to capture the failure.
Jason R. Coombs [Sun, 31 Aug 2014 19:02:42 +0000 (15:02 -0400)]
#22315: Add test to capture the failure.

10 years agoCorrect indent
Jason R. Coombs [Sun, 31 Aug 2014 19:00:47 +0000 (15:00 -0400)]
Correct indent

10 years agoRemove unused import
Jason R. Coombs [Sun, 31 Aug 2014 17:43:02 +0000 (13:43 -0400)]
Remove unused import

10 years agopytime.h: remove duplicated "#ifndef Py_LIMITED_API"
Victor Stinner [Sun, 31 Aug 2014 13:48:55 +0000 (15:48 +0200)]
pytime.h: remove duplicated "#ifndef Py_LIMITED_API"

10 years agoCloses #21527: Add default number of workers to ThreadPoolExecutor. (Claudiu Popa.)
Guido van Rossum [Tue, 2 Sep 2014 17:39:18 +0000 (10:39 -0700)]
Closes #21527: Add default number of workers to ThreadPoolExecutor. (Claudiu Popa.)

10 years ago(Merge 3.4) Closes #22258: Fix the the internal function set_inheritable() on
Victor Stinner [Tue, 2 Sep 2014 09:49:48 +0000 (11:49 +0200)]
(Merge 3.4) Closes #22258: Fix the the internal function set_inheritable() on
Illumos.  This platform exposes the function ioctl(FIOCLEX), but calling it
fails with errno is ENOTTY: "Inappropriate ioctl for device". set_inheritable()
now falls back to the slower fcntl() (F_GETFD and then F_SETFD).

10 years agosigma actually works in latex...
Benjamin Peterson [Mon, 1 Sep 2014 22:26:22 +0000 (18:26 -0400)]
sigma actually works in latex...

10 years agotry to pick a unicode char that latex understands
Benjamin Peterson [Mon, 1 Sep 2014 22:04:02 +0000 (18:04 -0400)]
try to pick a unicode char that latex understands

10 years agoa more universal unicode char example
Benjamin Peterson [Mon, 1 Sep 2014 21:59:18 +0000 (17:59 -0400)]
a more universal unicode char example

10 years agoRemove unused imports.
Berker Peksag [Mon, 1 Sep 2014 09:33:41 +0000 (12:33 +0300)]
Remove unused imports.

10 years agoIssue #19447: Suppress output of py_compile.compile().
Berker Peksag [Mon, 1 Sep 2014 09:30:17 +0000 (12:30 +0300)]
Issue #19447: Suppress output of py_compile.compile().

10 years agoIssue #22320: merge from 3.4
Ned Deily [Mon, 1 Sep 2014 07:08:00 +0000 (00:08 -0700)]
Issue #22320: merge from 3.4

10 years agoMerge with 3.4; Closes #22315
Jason R. Coombs [Sun, 31 Aug 2014 22:02:18 +0000 (18:02 -0400)]
Merge with 3.4; Closes #22315

10 years agomerge 3.4
Benjamin Peterson [Sun, 31 Aug 2014 21:22:27 +0000 (17:22 -0400)]
merge 3.4

10 years ago(Merge 3.4) asyncio, Tulip issue 205: Fix a race condition in
Victor Stinner [Sun, 31 Aug 2014 13:08:21 +0000 (15:08 +0200)]
(Merge 3.4) asyncio, Tulip issue 205: Fix a race condition in
BaseSelectorEventLoop.sock_connect()

There is a race condition in create_connection() used with wait_for() to have a
timeout. sock_connect() registers the file descriptor of the socket to be
notified of write event (if connect() raises BlockingIOError). When
create_connection() is cancelled with a TimeoutError, sock_connect() coroutine
gets the exception, but it doesn't unregister the file descriptor for write
event. create_connection() gets the TimeoutError and closes the socket.

If you call again create_connection(), the new socket will likely gets the same
file descriptor, which is still registered in the selector. When sock_connect()
calls add_writer(), it tries to modify the entry instead of creating a new one.

This issue was originally reported in the Trollius project, but the bug comes
from Tulip in fact (Trollius is based on Tulip):
https://bitbucket.org/enovance/trollius/issue/15/after-timeouterror-on-wait_for

This change fixes the race condition. It also makes sock_connect() more
reliable (and portable) is sock.connect() raises an InterruptedError.

10 years agoasyncio, Tulip issue 205: Fix a race condition in BaseSelectorEventLoop.sock_connect()
Victor Stinner [Sun, 31 Aug 2014 13:07:57 +0000 (15:07 +0200)]
asyncio, Tulip issue 205: Fix a race condition in BaseSelectorEventLoop.sock_connect()

There is a race condition in create_connection() used with wait_for() to have a
timeout. sock_connect() registers the file descriptor of the socket to be
notified of write event (if connect() raises BlockingIOError). When
create_connection() is cancelled with a TimeoutError, sock_connect() coroutine
gets the exception, but it doesn't unregister the file descriptor for write
event. create_connection() gets the TimeoutError and closes the socket.

If you call again create_connection(), the new socket will likely gets the same
file descriptor, which is still registered in the selector. When sock_connect()
calls add_writer(), it tries to modify the entry instead of creating a new one.

This issue was originally reported in the Trollius project, but the bug comes
from Tulip in fact (Trollius is based on Tulip):
https://bitbucket.org/enovance/trollius/issue/15/after-timeouterror-on-wait_for

This change fixes the race condition. It also makes sock_connect() more
reliable (and portable) is sock.connect() raises an InterruptedError.

10 years ago(Merge 3.4) Closes #22275: asyncio: enhance documentation of OS support
Victor Stinner [Sun, 31 Aug 2014 12:47:56 +0000 (14:47 +0200)]
(Merge 3.4) Closes #22275: asyncio: enhance documentation of OS support

10 years agoCloses #22275: asyncio: enhance documentation of OS support
Victor Stinner [Sun, 31 Aug 2014 12:47:37 +0000 (14:47 +0200)]
Closes #22275: asyncio: enhance documentation of OS support

10 years agomerge 3.4 (closes #22307)
Benjamin Peterson [Sun, 31 Aug 2014 01:06:00 +0000 (21:06 -0400)]
merge 3.4 (closes #22307)

10 years agogetuid() returns real process id not effective (closes #22307)
Benjamin Peterson [Sun, 31 Aug 2014 01:04:15 +0000 (21:04 -0400)]
getuid() returns real process id not effective (closes #22307)

10 years agoMerge #22215: have the smtplib 'quit' command reset the state.
R David Murray [Sat, 30 Aug 2014 20:55:45 +0000 (16:55 -0400)]
Merge #22215: have the smtplib 'quit' command reset the state.

10 years ago#22215: have the smtplib 'quit' command reset the state.
R David Murray [Sat, 30 Aug 2014 20:51:59 +0000 (16:51 -0400)]
#22215: have the smtplib 'quit' command reset the state.

Without this reset, starttls would fail if a connect/starttls was done after a
quit, because smtplib assumed the existing value of emspt_features was
accurate, but it gets reset when starttls completes (and the new value does
not contain the starttls capability, since tls is already started at that
point).  (There may be additional places where this lack of reset was an
issue as well.)

Patch by Milan Oberkirch.

10 years agoIssue #22098: ctypes' BigEndianStructure and LittleEndianStructure now define an...
Antoine Pitrou [Fri, 29 Aug 2014 22:37:18 +0000 (00:37 +0200)]
Issue #22098: ctypes' BigEndianStructure and LittleEndianStructure now define an empty __slots__ so that subclasses don't always get an instance dict.

Patch by Claudiu Popa.

10 years agoIssue #22185: Fix an occasional RuntimeError in threading.Condition.wait() caused...
Antoine Pitrou [Fri, 29 Aug 2014 21:27:33 +0000 (23:27 +0200)]
Issue #22185: Fix an occasional RuntimeError in threading.Condition.wait() caused by mutation of the waiters queue without holding the lock.

Patch by Doug Zongker.

10 years agoIssue #22185: Fix an occasional RuntimeError in threading.Condition.wait() caused...
Antoine Pitrou [Fri, 29 Aug 2014 21:26:36 +0000 (23:26 +0200)]
Issue #22185: Fix an occasional RuntimeError in threading.Condition.wait() caused by mutation of the waiters queue without holding the lock.

Patch by Doug Zongker.

10 years agoIssue #22043: Oops, fix perf_counter() on UNIX if no monotonic clock is
Victor Stinner [Fri, 29 Aug 2014 15:00:17 +0000 (17:00 +0200)]
Issue #22043: Oops, fix perf_counter() on UNIX if no monotonic clock is
available (unlikely)

10 years agoIssue #22043: Simplify time.perf_counter() on Windows
Victor Stinner [Fri, 29 Aug 2014 14:51:33 +0000 (16:51 +0200)]
Issue #22043: Simplify time.perf_counter() on Windows

QueryPerformanceFrequency() cannot fail on Windows XP and later according to
its documentation: raise an exception on error and drop the fallback to the
system clock.

10 years agoIssue #22043: _PyTime_Init() now checks if the system clock works.
Victor Stinner [Fri, 29 Aug 2014 14:31:59 +0000 (16:31 +0200)]
Issue #22043: _PyTime_Init() now checks if the system clock works.

Other changes:

* The whole _PyTime API is private (not defined if Py_LIMITED_API is set)
* _PyTime_gettimeofday_info() also returns -1 on error
* Simplify PyTime_gettimeofday(): only use clock_gettime(CLOCK_REALTIME) or
  gettimeofday() on UNIX. Don't fallback to ftime() or time() anymore.

10 years agoIssue #22287: On UNIX, _PyTime_gettimeofday() now uses
Victor Stinner [Fri, 29 Aug 2014 13:41:08 +0000 (15:41 +0200)]
Issue #22287: On UNIX, _PyTime_gettimeofday() now uses
clock_gettime(CLOCK_REALTIME) if available. As a side effect, Python now
depends on the librt library on Solaris and on Linux (only with glibc older
than 2.17).

10 years agoIssue #22182: Use e.args to unpack exceptions correctly in distutils.file_util.move_file.
Berker Peksag [Fri, 29 Aug 2014 04:09:25 +0000 (07:09 +0300)]
Issue #22182: Use e.args to unpack exceptions correctly in distutils.file_util.move_file.

Patch by Claudiu Popa.

10 years agoIssue #22182: Use e.args to unpack exceptions correctly in distutils.file_util.move_file.
Berker Peksag [Fri, 29 Aug 2014 04:07:35 +0000 (07:07 +0300)]
Issue #22182: Use e.args to unpack exceptions correctly in distutils.file_util.move_file.

Patch by Claudiu Popa.

10 years ago(Merge 3.4) asyncio, Tulip issue 201: Fix a race condition in wait_for()
Victor Stinner [Thu, 28 Aug 2014 09:19:46 +0000 (11:19 +0200)]
(Merge 3.4) asyncio, Tulip issue 201: Fix a race condition in wait_for()

Don't raise a TimeoutError if we reached the timeout and the future completed
in the same iteration of the event loop. A side effect of the bug is that
Queue.get() looses items.

10 years agoasyncio, Tulip issue 201: Fix a race condition in wait_for()
Victor Stinner [Thu, 28 Aug 2014 09:19:25 +0000 (11:19 +0200)]
asyncio, Tulip issue 201: Fix a race condition in wait_for()

Don't raise a TimeoutError if we reached the timeout and the future completed
in the same iteration of the event loop. A side effect of the bug is that
Queue.get() looses items.

10 years agoThe webbrowser module now uses subprocess's start_new_session=True rather
Gregory P. Smith [Wed, 27 Aug 2014 16:41:05 +0000 (09:41 -0700)]
The webbrowser module now uses subprocess's start_new_session=True rather
than a potentially risky preexec_fn=os.setsid call.

10 years agoThe webbrowser module now uses subprocess's start_new_session=True rather
Gregory P. Smith [Wed, 27 Aug 2014 16:34:38 +0000 (09:34 -0700)]
The webbrowser module now uses subprocess's start_new_session=True rather
than a potentially risky preexec_fn=os.setsid call.

10 years agoIssue #22042: Fix test_signal on Windows
Victor Stinner [Wed, 27 Aug 2014 12:02:36 +0000 (14:02 +0200)]
Issue #22042: Fix test_signal on Windows

10 years agoIssue #22042: signal.set_wakeup_fd(fd) now raises an exception if the file
Victor Stinner [Wed, 27 Aug 2014 10:59:44 +0000 (12:59 +0200)]
Issue #22042: signal.set_wakeup_fd(fd) now raises an exception if the file
descriptor is in blocking mode.

10 years agoMerge with 3.4
Terry Jan Reedy [Wed, 27 Aug 2014 05:58:57 +0000 (01:58 -0400)]
Merge with 3.4

10 years agoIssue #22065: Remove the now unsed configGUI menu parameter and arguments.
Terry Jan Reedy [Wed, 27 Aug 2014 05:58:40 +0000 (01:58 -0400)]
Issue #22065: Remove the now unsed configGUI menu parameter and arguments.

10 years agoMerge with 3.4
Terry Jan Reedy [Wed, 27 Aug 2014 05:44:13 +0000 (01:44 -0400)]
Merge with 3.4

10 years agoIssue #22065: Menus, unlike Menubottons, do not have a state option.
Terry Jan Reedy [Wed, 27 Aug 2014 05:43:50 +0000 (01:43 -0400)]
Issue #22065: Menus, unlike Menubottons, do not have a state option.
Since a new demo can now be loaded while one is running,
adjust loadfile to do the same cleanup as stopIt.

10 years agoMerge 3.4.
Stefan Krah [Tue, 26 Aug 2014 19:32:26 +0000 (21:32 +0200)]
Merge 3.4.

10 years agoIntroduce and check for MPD_VERSION_HEX for precise management of builds
Stefan Krah [Tue, 26 Aug 2014 19:31:47 +0000 (21:31 +0200)]
Introduce and check for MPD_VERSION_HEX for precise management of builds
with an external libmpdec.

10 years agoMerge 3.4.
Stefan Krah [Tue, 26 Aug 2014 18:47:32 +0000 (20:47 +0200)]
Merge 3.4.

10 years agoIssue #22090: Fix '%' formatting for infinities and NaNs.
Stefan Krah [Tue, 26 Aug 2014 18:46:49 +0000 (20:46 +0200)]
Issue #22090: Fix '%' formatting for infinities and NaNs.

10 years ago(Merge 3.4) Issue #22063: Fix asyncio documentation of socket and pipe
Victor Stinner [Mon, 25 Aug 2014 23:02:28 +0000 (01:02 +0200)]
(Merge 3.4) Issue #22063: Fix asyncio documentation of socket and pipe
operations regarding to non-blocking mode. The non-blocking mode does not
matter when using a ProactorEventLoop, only for SelectorEventLoop.

10 years agoIssue #22063: Fix asyncio documentation of socket and pipe operations regarding
Victor Stinner [Mon, 25 Aug 2014 23:01:59 +0000 (01:01 +0200)]
Issue #22063: Fix asyncio documentation of socket and pipe operations regarding
to non-blocking mode. The non-blocking mode does not matter when using a
ProactorEventLoop, only for SelectorEventLoop.

10 years ago(Merge 3.4) asyncio, tulip issue 203: Add _FlowControlMixin.get_write_buffer_limits()
Victor Stinner [Mon, 25 Aug 2014 22:23:23 +0000 (00:23 +0200)]
(Merge 3.4) asyncio, tulip issue 203: Add _FlowControlMixin.get_write_buffer_limits()
method

10 years agoasyncio, tulip issue 203: Add _FlowControlMixin.get_write_buffer_limits() method
Victor Stinner [Mon, 25 Aug 2014 22:22:28 +0000 (00:22 +0200)]
asyncio, tulip issue 203: Add _FlowControlMixin.get_write_buffer_limits() method

10 years ago(Merge 3.4) asyncio: sync with Tulip
Victor Stinner [Mon, 25 Aug 2014 21:22:54 +0000 (23:22 +0200)]
(Merge 3.4) asyncio: sync with Tulip

* PipeServer.close() now cancels the "accept pipe" future which cancels the
  overlapped operation.
* Fix _SelectorTransport.__repr__() if the transport was closed
* Fix debug log in BaseEventLoop.create_connection(): get the socket object
  from the transport because SSL transport closes the old socket and creates a
  new SSL socket object. Remove also the _SelectorSslTransport._rawsock
  attribute: it contained the closed socket (not very useful) and it was not
  used.
* Issue #22063: socket operations (sock_recv, sock_sendall, sock_connect,
  sock_accept) of the proactor event loop don't raise an exception in debug
  mode if the socket are in blocking mode. Overlapped operations also work on
  blocking sockets.
* Fix unit tests in debug mode: mock a non-blocking socket for socket
  operations which now raise an exception if the socket is blocking.
* _fatal_error() method of _UnixReadPipeTransport and _UnixWritePipeTransport
  now log all exceptions in debug mode
* Don't log expected errors in unit tests
* Tulip issue 200: _WaitHandleFuture._unregister_wait() now catchs and logs
  exceptions.
* Tulip issue 200: Log errors in debug mode instead of simply ignoring them.

10 years agoasyncio: sync with Tulip
Victor Stinner [Mon, 25 Aug 2014 21:20:52 +0000 (23:20 +0200)]
asyncio: sync with Tulip

* PipeServer.close() now cancels the "accept pipe" future which cancels the
  overlapped operation.
* Fix _SelectorTransport.__repr__() if the transport was closed
* Fix debug log in BaseEventLoop.create_connection(): get the socket object
  from the transport because SSL transport closes the old socket and creates a
  new SSL socket object. Remove also the _SelectorSslTransport._rawsock
  attribute: it contained the closed socket (not very useful) and it was not
  used.
* Issue #22063: socket operations (sock_recv, sock_sendall, sock_connect,
  sock_accept) of the proactor event loop don't raise an exception in debug
  mode if the socket are in blocking mode. Overlapped operations also work on
  blocking sockets.
* Fix unit tests in debug mode: mock a non-blocking socket for socket
  operations which now raise an exception if the socket is blocking.
* _fatal_error() method of _UnixReadPipeTransport and _UnixWritePipeTransport
  now log all exceptions in debug mode
* Don't log expected errors in unit tests
* Tulip issue 200: _WaitHandleFuture._unregister_wait() now catchs and logs
  exceptions.
* Tulip issue 200: Log errors in debug mode instead of simply ignoring them.

10 years ago(Merge 3.4) asyncio: update the doc
Victor Stinner [Mon, 25 Aug 2014 15:04:37 +0000 (17:04 +0200)]
(Merge 3.4) asyncio: update the doc

* dev: mention that the logging must be configured at DEBUG level
* streams: drain() has no more a strange return value, it's just
  a standard coroutine

10 years agoasyncio: update the doc
Victor Stinner [Mon, 25 Aug 2014 15:04:12 +0000 (17:04 +0200)]
asyncio: update the doc

* dev: mention that the logging must be configured at DEBUG level
* streams: drain() has no more a strange return value, it's just
  a standard coroutine

10 years agomerge 3.4 (#22265)
Benjamin Peterson [Sun, 24 Aug 2014 23:08:22 +0000 (18:08 -0500)]
merge 3.4 (#22265)

10 years agoallow test to work on implementations not using ref-counting (closes #22265)
Benjamin Peterson [Sun, 24 Aug 2014 23:07:28 +0000 (18:07 -0500)]
allow test to work on implementations not using ref-counting (closes #22265)

10 years agomerge 3.4 (#22267)
Benjamin Peterson [Sun, 24 Aug 2014 23:04:25 +0000 (18:04 -0500)]
merge 3.4 (#22267)

10 years agofix some test_weakref tests to not rely on ref-counting (closes #22267)
Benjamin Peterson [Sun, 24 Aug 2014 23:02:15 +0000 (18:02 -0500)]
fix some test_weakref tests to not rely on ref-counting (closes #22267)

10 years agoIssue #16808: inspect.stack() now returns a named tuple instead of a tuple.
Antoine Pitrou [Sun, 24 Aug 2014 14:50:28 +0000 (10:50 -0400)]
Issue #16808: inspect.stack() now returns a named tuple instead of a tuple.
Patch by Daniel Shahaf.

10 years agoIssue #22034: Improve handling of wrong argument types in posixpath.join().
Serhiy Storchaka [Sun, 24 Aug 2014 09:23:36 +0000 (12:23 +0300)]
Issue #22034: Improve handling of wrong argument types in posixpath.join().

10 years agoIssue #22034: Got rid of misleading error message for bytearray arguments in
Serhiy Storchaka [Sun, 24 Aug 2014 09:18:09 +0000 (12:18 +0300)]
Issue #22034: Got rid of misleading error message for bytearray arguments in
posixpath.join().

10 years agoIssue #22236: Tkinter tests now don't reuse default root window. New root
Serhiy Storchaka [Sun, 24 Aug 2014 06:10:58 +0000 (09:10 +0300)]
Issue #22236: Tkinter tests now don't reuse default root window.  New root
window is created for every test class.

Fixed Tkinter images copying operations in NoDefaultRoot mode.

Tcl command names generated for "after" callbacks now contains a name of
original function.

10 years agoIssue #22236: Tkinter tests now don't reuse default root window. New root
Serhiy Storchaka [Sun, 24 Aug 2014 06:07:47 +0000 (09:07 +0300)]
Issue #22236: Tkinter tests now don't reuse default root window.  New root
window is created for every test class.

Fixed Tkinter images copying operations in NoDefaultRoot mode.

Tcl command names generated for "after" callbacks now contains a name of
original function.

10 years agoMerge with 3.4
Terry Jan Reedy [Sun, 24 Aug 2014 02:22:07 +0000 (22:22 -0400)]
Merge with 3.4

10 years agoIssue #13540: add missing markup.
Terry Jan Reedy [Sun, 24 Aug 2014 02:21:47 +0000 (22:21 -0400)]
Issue #13540: add missing markup.

10 years agoIssue #21166: fix typo in comment
Ned Deily [Sun, 24 Aug 2014 01:10:49 +0000 (18:10 -0700)]
Issue #21166: fix typo in comment

10 years agoIssue #21166: fix typo in comment
Ned Deily [Sun, 24 Aug 2014 01:10:16 +0000 (18:10 -0700)]
Issue #21166: fix typo in comment

10 years agoMerge with 3.4
Terry Jan Reedy [Sat, 23 Aug 2014 23:29:58 +0000 (19:29 -0400)]
Merge with 3.4

10 years agoIssue #22243: fix except grammar in reference.
Terry Jan Reedy [Sat, 23 Aug 2014 23:29:47 +0000 (19:29 -0400)]
Issue #22243: fix except grammar in reference.

10 years agoMerge with 3.4
Terry Jan Reedy [Sat, 23 Aug 2014 22:29:02 +0000 (18:29 -0400)]
Merge with 3.4

10 years agoIssue #22232 (partial fix): update Universal newlines Glossary entry.
Terry Jan Reedy [Sat, 23 Aug 2014 22:28:44 +0000 (18:28 -0400)]
Issue #22232 (partial fix): update Universal newlines Glossary entry.

10 years agoMerge with 3.4
Terry Jan Reedy [Sat, 23 Aug 2014 22:24:09 +0000 (18:24 -0400)]
Merge with 3.4

10 years agoMerge heads.
Terry Jan Reedy [Sat, 23 Aug 2014 22:23:50 +0000 (18:23 -0400)]
Merge heads.

10 years agoMerge heads.
Terry Jan Reedy [Sat, 23 Aug 2014 22:22:43 +0000 (18:22 -0400)]
Merge heads.

10 years agoMerge PEP 466 What's New in 2.7 changes from 3.4
Nick Coghlan [Sat, 23 Aug 2014 04:48:22 +0000 (14:48 +1000)]
Merge PEP 466 What's New in 2.7 changes from 3.4

10 years agoForward port PEP 466 What's New in 2.7 updates
Nick Coghlan [Sat, 23 Aug 2014 04:47:47 +0000 (14:47 +1000)]
Forward port PEP 466 What's New in 2.7 updates

10 years agoIssue #2527: Add a *globals* argument to timeit functions, in order to override the...
Antoine Pitrou [Sat, 23 Aug 2014 03:13:50 +0000 (23:13 -0400)]
Issue #2527: Add a *globals* argument to timeit functions, in order to override the globals namespace in which the timed code is executed.

Patch by Ben Roberts.

10 years agoIssue #21166: merge from 3.4
Ned Deily [Fri, 22 Aug 2014 20:36:30 +0000 (13:36 -0700)]
Issue #21166: merge from 3.4

10 years agoIssue #21166: Prevent possible segfaults and other random failures of
Ned Deily [Fri, 22 Aug 2014 20:32:49 +0000 (13:32 -0700)]
Issue #21166: Prevent possible segfaults and other random failures of
python --generate-posix-vars in pybuilddir.txt build target by ensuring
that pybuilddir.txt is always regenerated when configure is run and
that the newly built skeleton python does not inadvertently import
modules from previously installed instances.

10 years agoIssue #20152: Port pyexpat to Argument Clinic.
Brett Cannon [Fri, 22 Aug 2014 18:23:20 +0000 (14:23 -0400)]
Issue #20152: Port pyexpat to Argument Clinic.

Could not emit an external file as pyexpat has a conditionally built
method which Clinic won't hide otherwise.

10 years agoAdd a missing Argument Clinic file
Brett Cannon [Fri, 22 Aug 2014 18:08:46 +0000 (14:08 -0400)]
Add a missing Argument Clinic file

10 years agoIssue #20152: Port the pwd module to Argument Clinic.
Brett Cannon [Fri, 22 Aug 2014 18:03:51 +0000 (14:03 -0400)]
Issue #20152: Port the pwd module to Argument Clinic.