]> git.ipfire.org Git - thirdparty/Python/cpython.git/log
thirdparty/Python/cpython.git
5 years agoMake 2.7.18rc1 release notes.
Benjamin Peterson [Sat, 4 Apr 2020 16:53:42 +0000 (11:53 -0500)] 
Make 2.7.18rc1 release notes.

5 years agoUpdate macOS installer build for 2.7.18 end-of-life. (GH-19352)
Ned Deily [Sat, 4 Apr 2020 00:34:39 +0000 (20:34 -0400)] 
Update macOS installer build for 2.7.18 end-of-life. (GH-19352)

5 years ago[2.7] closes bpo-40125: Update multissltests.py to use OpenSSL 1.1.1f. (GH-19251)
Benjamin Peterson [Wed, 1 Apr 2020 01:52:23 +0000 (20:52 -0500)] 
[2.7] closes bpo-40125: Update multissltests.py to use OpenSSL 1.1.1f. (GH-19251)

(cherry picked from commit cd16661f903153ecac55f190ed682e576c5deb24)

5 years ago[2.7] closes bpo-38576: Disallow control characters in hostnames in http.client....
Matěj Cepl [Thu, 19 Mar 2020 01:35:44 +0000 (02:35 +0100)] 
[2.7] closes bpo-38576: Disallow control characters in hostnames in http.client. (GH-19052)

Add host validation for control characters for more
CVE-2019-18348 protection.
(cherry picked from commit 83fc70159b24)

Co-authored-by: Ashwin Ramaswami <aramaswamis@gmail.com>
6 years agoDoc: Change Python 2 status to EOL. (GH-17885)
Miss Islington (bot) [Tue, 7 Jan 2020 06:56:57 +0000 (22:56 -0800)] 
Doc: Change Python 2 status to EOL. (GH-17885)

(cherry picked from commit f4800b8ed3dbe15a0078869a836d968ab3362b8c)

Co-authored-by: Inada Naoki <songofacandy@gmail.com>
6 years agobpo-27973 - Use test.support.temp_dir instead of NamedTemporaryFile for the (#17774)
Senthil Kumaran [Sat, 4 Jan 2020 02:14:18 +0000 (18:14 -0800)] 
bpo-27973 - Use test.support.temp_dir instead of NamedTemporaryFile for the (#17774)

desired behavior under windows platform.

Suggestion by David Bolen

6 years agoUpdate copyright year in macOS installer license copy (GH-17806)
Miss Islington (bot) [Fri, 3 Jan 2020 03:44:03 +0000 (19:44 -0800)] 
Update copyright year in macOS installer license copy (GH-17806)

(cherry picked from commit 32f1443aa98db769d87db497b45bd0dcb732445b)

Co-authored-by: Ned Deily <nad@python.org>
6 years ago[2.7] Bring Python into the next decade. (GH-17805)
Benjamin Peterson [Fri, 3 Jan 2020 03:10:06 +0000 (19:10 -0800)] 
[2.7] Bring Python into the next decade. (GH-17805)

(cherry picked from commit 946b29ea0b3b386ed05e87e60b8617c9dc19cd53)

Co-authored-by: Benjamin Peterson <benjamin@python.org>
6 years ago[2.7] bpo-27973 - Fix for urllib.urlretrieve() failing on second ftp transfer (#1040)
Senthil Kumaran [Tue, 31 Dec 2019 05:14:56 +0000 (21:14 -0800)] 
[2.7] bpo-27973 - Fix for urllib.urlretrieve() failing on second ftp transfer (#1040)

* bpo-27973: Fix urllib.urlretrieve failing on subsequent ftp transfers from the same host.

* bpo-35411: Skip test_urllibnet FTP tests on Travis CI.

6 years ago[2.7] Minor C API documentation improvements. (GH-17699)
Benjamin Peterson [Wed, 25 Dec 2019 04:34:38 +0000 (22:34 -0600)] 
[2.7] Minor C API documentation improvements. (GH-17699)

(cherry picked from commit 5c7ed7550ec2da16d7679e538fcd7c1a5631811f)

Co-authored-by: William Ayd <william.ayd@icloud.com>
6 years agobpo-38295: prevent test_relative_path of test_py_compile failure on macOS Catalina...
Miss Islington (bot) [Tue, 17 Dec 2019 09:16:33 +0000 (01:16 -0800)] 
bpo-38295: prevent test_relative_path of test_py_compile failure on macOS Catalina (GH-17636)

(cherry picked from commit bf3aa1060a29a05813abbe877193af16e3e7131e)

Co-authored-by: Ned Deily <nad@python.org>
6 years agobpo-38730: Replace strncpy in import.c with memcpy. (GH-17633)
Benjamin Peterson [Tue, 17 Dec 2019 00:39:57 +0000 (16:39 -0800)] 
bpo-38730: Replace strncpy in import.c with memcpy. (GH-17633)

In all these cases, we know the exact length we want copied, so memcpy is the right function to use.

6 years agobpo-39035: travis: Don't use beta group (GH-17605)
Inada Naoki [Sat, 14 Dec 2019 14:02:55 +0000 (23:02 +0900)] 
bpo-39035: travis: Don't use beta group (GH-17605)

6 years ago[2.7] bpo-38945: UU Encoding: Don't let newline in filename corrupt the output format... 17574/head
Matthew Rollings [Tue, 3 Dec 2019 18:18:52 +0000 (18:18 +0000)] 
[2.7] bpo-38945: UU Encoding: Don't let newline in filename corrupt the output format (GH-17418). (#17452)

(cherry picked from commit a62ad4730c9b575f140f24074656c0257c86a09a)

Co-authored-by: Matthew Rollings <1211162+stealthcopter@users.noreply.github.com>
6 years agodocument threading.Lock.locked() (GH-17427)
Miss Islington (bot) [Sun, 1 Dec 2019 20:12:09 +0000 (12:12 -0800)] 
document threading.Lock.locked() (GH-17427)

(cherry picked from commit fdafa1d0ed0a8930b52ee81e57c931cc4d5c2388)

Co-authored-by: idomic <michael.ido@gmail.com>
6 years agobpo-38804: Fix REDoS in http.cookiejar (GH-17157) (GH-17345)
Victor Stinner [Sun, 24 Nov 2019 15:49:23 +0000 (16:49 +0100)] 
bpo-38804: Fix REDoS in http.cookiejar (GH-17157) (GH-17345)

The regex http.cookiejar.LOOSE_HTTP_DATE_RE was vulnerable to regular
expression denial of service (REDoS).

LOOSE_HTTP_DATE_RE.match is called when using http.cookiejar.CookieJar
to parse Set-Cookie headers returned by a server.
Processing a response from a malicious HTTP server can lead to extreme
CPU usage and execution will be blocked for a long time.

The regex contained multiple overlapping \s* capture groups.
Ignoring the ?-optional capture groups the regex could be simplified to

    \d+-\w+-\d+(\s*\s*\s*)$

Therefore, a long sequence of spaces can trigger bad performance.

Matching a malicious string such as

    LOOSE_HTTP_DATE_RE.match("1-c-1" + (" " * 2000) + "!")

caused catastrophic backtracking.

The fix removes ambiguity about which \s* should match a particular
space.

You can create a malicious server which responds with Set-Cookie headers
to attack all python programs which access it e.g.

    from http.server import BaseHTTPRequestHandler, HTTPServer

    def make_set_cookie_value(n_spaces):
        spaces = " " * n_spaces
        expiry = f"1-c-1{spaces}!"
        return f"b;Expires={expiry}"

    class Handler(BaseHTTPRequestHandler):
        def do_GET(self):
            self.log_request(204)
            self.send_response_only(204)  # Don't bother sending Server and Date
            n_spaces = (
                int(self.path[1:])  # Can GET e.g. /100 to test shorter sequences
                if len(self.path) > 1 else
                65506  # Max header line length 65536
            )
            value = make_set_cookie_value(n_spaces)
            for i in range(99):  # Not necessary, but we can have up to 100 header lines
                self.send_header("Set-Cookie", value)
            self.end_headers()

    if __name__ == "__main__":
        HTTPServer(("", 44020), Handler).serve_forever()

This server returns 99 Set-Cookie headers. Each has 65506 spaces.
Extracting the cookies will pretty much never complete.

Vulnerable client using the example at the bottom of
https://docs.python.org/3/library/http.cookiejar.html :

    import http.cookiejar, urllib.request
    cj = http.cookiejar.CookieJar()
    opener = urllib.request.build_opener(urllib.request.HTTPCookieProcessor(cj))
    r = opener.open("http://localhost:44020/")

The popular requests library was also vulnerable without any additional
options (as it uses http.cookiejar by default):

    import requests
    requests.get("http://localhost:44020/")

* Regression test for http.cookiejar REDoS

If we regress, this test will take a very long time.

* Improve performance of http.cookiejar.ISO_DATE_RE

A string like

"444444" + (" " * 2000) + "A"

could cause poor performance due to the 2 overlapping \s* groups,
although this is not as serious as the REDoS in LOOSE_HTTP_DATE_RE was.

(cherry picked from commit 1b779bfb8593739b11cbb988ef82a883ec9d077e)

6 years agobpo-38730: Remove usage of stpncpy as it's not supported on MSVC 2008. (GH-17081)
Benjamin Peterson [Thu, 7 Nov 2019 15:27:03 +0000 (07:27 -0800)] 
bpo-38730: Remove usage of stpncpy as it's not supported on MSVC 2008. (GH-17081)

6 years ago[2.7] bpo-38730: Fix -Wstringop-truncation warnings. (GH-17075)
Benjamin Peterson [Thu, 7 Nov 2019 15:06:28 +0000 (07:06 -0800)] 
[2.7] bpo-38730: Fix -Wstringop-truncation warnings. (GH-17075)

6 years agobpo-37731: Squish another _POSIX_C_SOURCE redefinition problem in expat. (GH-17077)
Benjamin Peterson [Thu, 7 Nov 2019 05:29:43 +0000 (21:29 -0800)] 
bpo-37731: Squish another _POSIX_C_SOURCE redefinition problem in expat. (GH-17077)

6 years agobpo-37731: Reorder includes in xmltok.c to avoid redefinition of _POSIX_C_SOURCE...
Miss Islington (bot) [Thu, 7 Nov 2019 05:10:05 +0000 (21:10 -0800)] 
bpo-37731: Reorder includes in xmltok.c to avoid redefinition of _POSIX_C_SOURCE (GH-16733)

(cherry picked from commit 8177404d520e81f16324a900f093adf3856d33f8)

Co-authored-by: Pablo Galindo <Pablogsal@gmail.com>
6 years agobpo-38557: Improve documentation for list and tuple C API. (GH-16925)
Miss Skeleton (bot) [Sat, 26 Oct 2019 20:04:13 +0000 (13:04 -0700)] 
bpo-38557: Improve documentation for list and tuple C API. (GH-16925)

(cherry picked from commit d898d20e8c228229eb68e545f544db13f246f216)

Co-authored-by: Serhiy Storchaka <storchaka@gmail.com>
6 years ago[2.7] bpo-38535: Fix positions for AST nodes for calls without arguments in decorator...
Serhiy Storchaka [Sat, 26 Oct 2019 14:30:30 +0000 (17:30 +0300)] 
[2.7] bpo-38535: Fix positions for AST nodes for calls without arguments in decorators. (GH-16861). (GH-16931)

(cherry picked from commit 26ae9f6d3d755734c9f371b9356325afe5764813)

6 years agoUpdate URL in macOS installer copy of license (GH-16905)
Miss Skeleton (bot) [Wed, 23 Oct 2019 20:15:56 +0000 (13:15 -0700)] 
Update URL in macOS installer copy of license (GH-16905)

(cherry picked from commit 01659ca62c4508518478a74615ac91c0009427ad)

Co-authored-by: Ned Deily <nad@python.org>
6 years agobpo-37025: AddRefActCtx() shouldn't be checked for failure (GH-16897)
Zackery Spytz [Wed, 23 Oct 2019 18:15:55 +0000 (12:15 -0600)] 
bpo-37025: AddRefActCtx() shouldn't be checked for failure (GH-16897)

AddRefActCtx() does not return a value.

6 years agoFix Zope URL (GH-16880)
Miss Skeleton (bot) [Tue, 22 Oct 2019 09:48:33 +0000 (02:48 -0700)] 
Fix Zope URL (GH-16880)

(cherry picked from commit dfe726b1ace03f206f45253b93ed7610473ae20f)

Co-authored-by: Kyle Stanley <aeros167@gmail.com>
6 years ago[2.7] bpo-38540: Fix possible leak in PyArg_Parse for "es#" and "et#". (GH-16869...
Serhiy Storchaka [Mon, 21 Oct 2019 18:40:30 +0000 (21:40 +0300)] 
[2.7] bpo-38540: Fix possible leak in PyArg_Parse for "es#" and "et#". (GH-16869). (GH-16877)

(cherry picked from commit 5bc6a7c06eda20ba131ecba6752be0506d310181)

6 years agoWork around Path.glob() issue when creating nuget package (GH-16855)
Steve Dower [Sun, 20 Oct 2019 01:25:35 +0000 (18:25 -0700)] 
Work around Path.glob() issue when creating nuget package (GH-16855)

6 years ago2.2.17+
Benjamin Peterson [Sat, 19 Oct 2019 20:03:22 +0000 (13:03 -0700)] 
2.2.17+

6 years agoEmpty blurb file for 2.7.17. v2.7.17
Benjamin Peterson [Sat, 19 Oct 2019 18:38:44 +0000 (11:38 -0700)] 
Empty blurb file for 2.7.17.

6 years agoBump version to 2.7.17 final.
Benjamin Peterson [Sat, 19 Oct 2019 18:37:52 +0000 (11:37 -0700)] 
Bump version to 2.7.17 final.

6 years agoUpdate doc switcher list for 3.8.0 (GH-16809)
Miss Islington (bot) [Sat, 19 Oct 2019 17:52:07 +0000 (10:52 -0700)] 
Update doc switcher list for 3.8.0 (GH-16809)

(cherry picked from commit 3f36043db22361500f52634f2b8de49dde0e7da9)

Co-authored-by: Ned Deily <nad@python.org>
6 years agoUpdate build docs for macOS (GH-16844)
Ned Deily [Sat, 19 Oct 2019 09:35:44 +0000 (05:35 -0400)] 
Update build docs for macOS (GH-16844)

6 years agobpo-32758: Warn that ast.parse() and ast.literal_eval() can segfault the interpreter...
Ashley Whetter [Fri, 18 Oct 2019 08:00:22 +0000 (01:00 -0700)] 
bpo-32758: Warn that ast.parse() and ast.literal_eval() can segfault the interpreter (GH-5960) (GH-16565)

(cherry picked from commit 7a7f100eb352d08938ee0f5ba59c18f56dc4a7b5)

Co-authored-by: Brett Cannon <brettcannon@users.noreply.github.com>
6 years agobpo-32758: Warn that compile() can crash when compiling to an AST object (GH-6043...
Ashley Whetter [Fri, 18 Oct 2019 08:00:03 +0000 (01:00 -0700)] 
bpo-32758: Warn that compile() can crash when compiling to an AST object (GH-6043) (GH-16566)

(cherry picked from commit f7a6ff6fcab32a53f262ba3f8a072c27afc330d7)

Co-authored-by: Brett Cannon <brettcannon@users.noreply.github.com>
6 years agoDoc: 3.8 is now stable. (GH-16790) (GH-16794)
Miss Islington (bot) [Mon, 14 Oct 2019 22:22:18 +0000 (15:22 -0700)] 
Doc: 3.8 is now stable. (GH-16790) (GH-16794)

(cherry picked from commit 4504b4500d2a1a80c26b27b0bfff8b624d5ce06c)

Co-authored-by: Julien Palard <julien@palard.fr>
6 years ago[2.7] Update macOS installer display files for 2.7.17 (GH-16768)
Ned Deily [Mon, 14 Oct 2019 08:39:00 +0000 (04:39 -0400)] 
[2.7] Update macOS installer display files for 2.7.17 (GH-16768)

6 years ago[2.7] bpo-31036: Allow sphinx and blurb to be found automatically (GH-16638)
Benjamin Peterson [Tue, 8 Oct 2019 03:57:05 +0000 (20:57 -0700)] 
[2.7] bpo-31036: Allow sphinx and blurb to be found automatically (GH-16638)

Rather than requiring the path to blurb and/or sphinx-build to be specified to the make rule, enhance the Doc/Makefile to look for each first in a virtual environment created by make venv and, if not found, look on the normal process PATH. This allows the Doc/Makefile to take advantage of an installed spinx-build or blurb and, thus, do the right thing most of the time. Also, make the directory for the venv be configurable and document the `make venv` target..
(cherry picked from commit 590665c399fc4aa3c4a9f8e7104d43a02e9f3a0c)

Co-authored-by: Ned Deily <nad@python.org>
6 years agobpo-35036: Remove empty log line in the suspicious.py tool (GH-10024)
Miss Islington (bot) [Tue, 8 Oct 2019 03:43:53 +0000 (20:43 -0700)] 
bpo-35036: Remove empty log line in the suspicious.py tool (GH-10024)

Previous to commit ee171a2 the logline was working because of self.info() (now
deprecated) defaults to an empty message.
(cherry picked from commit c3f52a59ce8406d9e59253ad4621e4749abdaeef)

Co-authored-by: Xtreak <tirkarthi@users.noreply.github.com>
6 years agobpo-31589 : Build PDF using xelatex for better UTF8 support. (GH-3940)
Miss Islington (bot) [Tue, 8 Oct 2019 03:42:51 +0000 (20:42 -0700)] 
bpo-31589 : Build PDF using xelatex for better UTF8 support. (GH-3940)

Also addresses doc build failures documented in bpo-32200.
(cherry picked from commit 7324b5ce8e7c031a0a3832a6a8d7c639111ae0ff)

Co-authored-by: Julien Palard <julien@palard.fr>
6 years ago[2.7] Stop using deprecated logging API in Sphinx suspicious checker (GH-16635)
Benjamin Peterson [Tue, 8 Oct 2019 03:37:45 +0000 (20:37 -0700)] 
[2.7] Stop using deprecated logging API in Sphinx suspicious checker (GH-16635)

(cherry picked from commit ee171a26c1169abfae534b08acc0d95c6e45a22a)

Co-authored-by: Pablo Galindo <Pablogsal@gmail.com>
6 years agoUpdate macOS installer displays for 2.7.17rc1 (#16634) v2.7.17rc1
Ned Deily [Tue, 8 Oct 2019 02:13:04 +0000 (22:13 -0400)] 
Update macOS installer displays for 2.7.17rc1 (#16634)

6 years agoBump version to 2.7.17rc1.
Benjamin Peterson [Tue, 8 Oct 2019 02:03:32 +0000 (19:03 -0700)] 
Bump version to 2.7.17rc1.

6 years agoRoll up news for 2.7.17rc1.
Benjamin Peterson [Tue, 8 Oct 2019 02:01:18 +0000 (19:01 -0700)] 
Roll up news for 2.7.17rc1.

6 years ago[2.7] bpo-38216, bpo-36274: Allow subclasses to separately override validation and...
Jason R. Coombs [Tue, 8 Oct 2019 02:00:02 +0000 (22:00 -0400)] 
[2.7] bpo-38216, bpo-36274: Allow subclasses to separately override validation and encoding behavior (GH-16476)

Backporting this change, I observe a couple of things:

1. The _encode_request call is no longer meaningful because the request construction will implicitly encode the request using the default encoding when the format string is used (request = '%s %s %s'...). In order to keep the code as consistent as possible, I decided to include the call as a pass-through. I'd be just as happy to remove it entirely, but I'll leave that up to the reviewer to decide. It's okay that this functionality is disabled on Python 2 because this functionality was mainly around bpo-36274, which was mainly a concern with the transition to Python 3.
2. Because _encode_request is no longer meaningful, neither is the test for it, so I've removed that test. Therefore, the meaningful part of this test is that for bpo-38216, adding a (underscore-protected) hook to customize/disable validation.

(cherry picked from commit 7774d7831e8809795c64ce27f7df52674581d298)

Co-authored-by: Jason R. Coombs <jaraco@jaraco.com>
6 years ago[2.7] bpo-37664: Update ensurepip bundled wheels, again (GH-16633)
Benjamin Peterson [Tue, 8 Oct 2019 01:54:05 +0000 (18:54 -0700)] 
[2.7] bpo-37664: Update ensurepip bundled wheels, again (GH-16633)

(cherry picked from commit 10c452b894d95fed06056fe11e8fe8e1a2a60040)

Co-authored-by: Pradyun Gedam <pradyunsg@gmail.com>
6 years agobpo-38106: Fix race in pthread PyThread_release_lock() (GH-16047)
Kirill Smelkov [Thu, 3 Oct 2019 07:06:52 +0000 (10:06 +0300)] 
bpo-38106: Fix race in pthread PyThread_release_lock() (GH-16047)

Fix race in PyThread_release_lock that was leading to memory corruption and
deadlocks. The fix applies to POSIX systems where Python locks are implemented
with mutex and condition variable because POSIX semaphores are either not
provided, or are known to be broken. One particular example of such system is
macOS.

On Darwin, even though this is considered as POSIX, Python uses
mutex+condition variable to implement its lock, and, as of 2019-08-28, Py2.7
implementation, even though similar issue was fixed for Py3 in 2012, contains
synchronization bug: the condition is signalled after mutex unlock while the
correct protocol is to signal condition from under mutex:

  https://github.com/python/cpython/blob/v2.7.16-127-g0229b56d8c0/Python/thread_pthread.h#L486-L506
  https://github.com/python/cpython/commit/187aa545165d (py3 fix)

PyPy has the same bug for both pypy2 and pypy3:

  https://bitbucket.org/pypy/pypy/src/578667b3fef9/rpython/translator/c/src/thread_pthread.c#lines-443:465
  https://bitbucket.org/pypy/pypy/src/5b42890d48c3/rpython/translator/c/src/thread_pthread.c#lines-443:465

Signalling condition outside of corresponding mutex is considered OK by
POSIX, but in Python context it can lead to at least memory corruption if we
consider the whole lifetime of python level lock. For example the following
logical scenario:

      T1                                          T2

  sema = Lock()
  sema.acquire()

                                              sema.release()

  sema.acquire()
  free(sema)

  ...

can translate to the next C-level calls:

      T1                                          T2

  # sema = Lock()
  sema = malloc(...)
  sema.locked = 0
  pthread_mutex_init(&sema.mut)
  pthread_cond_init (&sema.lock_released)

  # sema.acquire()
  pthread_mutex_lock(&sema.mut)
  # sees sema.locked == 0
  sema.locked = 1
  pthread_mutex_unlock(&sema.mut)

                                              # sema.release()
                                              pthread_mutex_lock(&sema.mut)
                                              sema.locked = 0
                                              pthread_mutex_unlock(&sema.mut)

                      # OS scheduler gets in and relinquishes control from T2
                      # to another process
                                              ...

  # second sema.acquire()
  pthread_mutex_lock(&sema.mut)
  # sees sema.locked == 0
  sema.locked = 1
  pthread_mutex_unlock(&sema.mut)

  # free(sema)
  pthread_mutex_destroy(&sema.mut)
  pthread_cond_destroy (&sema.lock_released)
  free(sema)

  # ...
  e.g. malloc() which returns memory where sema was

                                              ...
                      # OS scheduler returns control to T2
                      # sema.release() continues
                      #
                      # BUT sema was already freed and writing to anywhere
                      # inside sema block CORRUPTS MEMORY. In particular if
                      # _another_ python-level lock was allocated where sema
                      # block was, writing into the memory can have effect on
                      # further synchronization correctness and in particular
                      # lead to deadlock on lock that was next allocated.
                                              pthread_cond_signal(&sema.lock_released)

Note that T2.pthread_cond_signal(&sema.lock_released) CORRUPTS MEMORY as it
is called when sema memory was already freed and is potentially
reallocated for another object.

The fix is to move pthread_cond_signal to be done under corresponding mutex:

  # sema.release()
  pthread_mutex_lock(&sema.mut)
  sema.locked = 0
  pthread_cond_signal(&sema.lock_released)
  pthread_mutex_unlock(&sema.mut)

To do so this patch cherry-picks thread_pthread.h part of the following 3.2 commit:

commit 187aa545165d8d5eac222ecce29c8a77e0282dd4
Author: Kristján Valur Jónsson <kristjan@ccpgames.com>
Date:   Tue Jun 5 22:17:42 2012 +0000

    Signal condition variables with the mutex held.  Destroy condition variables
    before their mutexes.

 Python/ceval_gil.h      |  9 +++++----
 Python/thread_pthread.h | 15 +++++++++------
 2 files changed, 14 insertions(+), 10 deletions(-)

(ceval_gil.h is Python3 specific and does not apply to Python2.7)

The bug was there since 1994 - since at least [1]. It was discussed in 2001
with original code author[2], but the code was still considered to be
race-free. In 2010 the place where pthread_cond_signal should be - before or
after pthread_mutex_unlock - was discussed with the rationale to avoid
threads bouncing[3,4,5], and in 2012 pthread_cond_signal was moved to be
called from under mutex, but only for CPython3[6,7].

In 2019 the bug was (re-)discovered while testing Pygolang[8] on macOS with
CPython2 and PyPy2 and PyPy3.

[1] https://github.com/python/cpython/commit/2c8cb9f3d240
[2] https://bugs.python.org/issue433625
[3] https://bugs.python.org/issue8299#msg103224
[4] https://bugs.python.org/issue8410#msg103313
[5] https://bugs.python.org/issue8411#msg113301
[6] https://bugs.python.org/issue15038#msg163187
[7] https://github.com/python/cpython/commit/187aa545165d
[8] https://pypi.org/project/pygolang

(cherry picked from commit 187aa545165d8d5eac222ecce29c8a77e0282dd4)

Co-Authored-By: Kristján Valur Jónsson <kristjan@ccpgames.com>
6 years ago[2.7] bpo-38338, test.pythoninfo: add more ssl infos (GH-16543)
Victor Stinner [Wed, 2 Oct 2019 16:36:32 +0000 (18:36 +0200)] 
[2.7] bpo-38338, test.pythoninfo: add more ssl infos (GH-16543)

test.pythoninfo now logs environment variables used by OpenSSL and
Python ssl modules, and logs attributes of 3 SSL contexts
(SSLContext, default HTTPS context, stdlib context).

(cherry picked from commit 1df1c2f8df53d005ff47af81aa02c58752b84e20)

6 years ago[2.7] bpo-38243: Escape the server title of DocXMLRPCServer (GH-16447)
Dong-hee Na [Tue, 1 Oct 2019 10:58:01 +0000 (19:58 +0900)] 
[2.7] bpo-38243: Escape the server title of DocXMLRPCServer (GH-16447)

Escape the server title of DocXMLRPCServer.DocXMLRPCServer
when rendering the document page as HTML.

6 years ago[2.7] bpo-38301: In Solaris family, we must be sure to use '-D_REENTRANT' (GH-16446...
Jesús Cea [Sat, 28 Sep 2019 03:09:24 +0000 (05:09 +0200)] 
[2.7] bpo-38301: In Solaris family, we must be sure to use '-D_REENTRANT' (GH-16446). (#16454)

(cherry picked from commit 52d1b86bde2b772a76919c76991c326384954bf1)

Co-authored-by: Jesús Cea <jcea@jcea.es>
6 years agobpo-38174 follow up: Remove loadlibrary.c from VS9.0. (GH-16411)
Benjamin Peterson [Thu, 26 Sep 2019 05:07:09 +0000 (22:07 -0700)] 
bpo-38174 follow up: Remove loadlibrary.c from VS9.0. (GH-16411)

6 years ago[2.7] closes bpo-38174: Update vendored expat library to 2.2.8. (GH-16408)
Benjamin Peterson [Thu, 26 Sep 2019 04:49:04 +0000 (21:49 -0700)] 
[2.7] closes bpo-38174: Update vendored expat library to 2.2.8. (GH-16408)

Fixes CVE-2019-15903. See full changelog at https://github.com/libexpat/libexpat/blob/R_2_2_8/expat/Changes..
(cherry picked from commit 52b940803860e37bcc3f6096b2d24e7c20a0e807)

Co-authored-by: Benjamin Peterson <benjamin@python.org>
6 years agocloses bpo-38253: Fix typo of Py_SET_ERANGE_IF_OVERFLOW in pyport.h. (GH-16230)
Miss Islington (bot) [Tue, 24 Sep 2019 02:31:25 +0000 (19:31 -0700)] 
closes bpo-38253: Fix typo of Py_SET_ERANGE_IF_OVERFLOW in pyport.h. (GH-16230)

(cherry picked from commit 4346bad3321699d49a45e3446270b57726ab5c8f)

Co-authored-by: Hai Shi <shihai1992@gmail.com>
6 years agobpo-37904: Edition on python tutorial - section 4 (GH-16169) (GH-16236)
Miss Islington (bot) [Wed, 18 Sep 2019 10:36:57 +0000 (03:36 -0700)] 
bpo-37904: Edition on python tutorial - section 4 (GH-16169) (GH-16236)

A little change on first paragraph of python tutorial to be more clearly

https://bugs.python.org/issue37904

Automerge-Triggered-By: @ericvsmith
(cherry picked from commit b57481318e3e3cbacd398b898f9849ec8f2d7eec)

Co-authored-by: Diego Alberto Barriga Martínez <diegobarriga@protonmail.com>
6 years ago[2.7] bpo-38175: Fix a memory leak in comparison of sqlite3.Row objects. (GH-16155...
Serhiy Storchaka [Tue, 17 Sep 2019 06:56:27 +0000 (09:56 +0300)] 
[2.7] bpo-38175: Fix a memory leak in comparison of sqlite3.Row objects. (GH-16155). (GH-16215)

(cherry picked from commit 8debfa50407107ff2329d01081cdc12d359f1d12)

6 years agobpo-33936: Don't call obsolete init methods with OpenSSL 1.1.0+ (GH-16140)
Miss Islington (bot) [Mon, 16 Sep 2019 19:48:21 +0000 (12:48 -0700)] 
bpo-33936: Don't call obsolete init methods with OpenSSL 1.1.0+ (GH-16140)

``OPENSSL_VERSION_1_1`` was never defined in ``_hashopenssl.c``.

https://bugs.python.org/issue33936
(cherry picked from commit 724f1a57231f9287c37255adf0e4364d12cf693d)

Co-authored-by: Christian Heimes <christian@python.org>
6 years agobpo-38117: Updates bundled OpenSSL to 1.0.2t (GH-16178)
Steve Dower [Mon, 16 Sep 2019 12:07:40 +0000 (13:07 +0100)] 
bpo-38117: Updates bundled OpenSSL to 1.0.2t (GH-16178)

6 years ago[2.7] bpo-38168: Fix a possbile refleak in setint() of mmapmodule.c (GH-16136) (GH...
Xiang Zhang [Mon, 16 Sep 2019 07:07:32 +0000 (15:07 +0800)] 
[2.7] bpo-38168: Fix a possbile refleak in setint() of mmapmodule.c (GH-16136) (GH-16176)

(cherry picked from commit 56a4514)

Co-authored-by: Hai Shi shihai1992@gmail.com
https://bugs.python.org/issue38168

6 years agobpo-38117: Updated OpenSSL to 1.0.2t in macOS installer for 2.7.x. (GH-16171)
Ned Deily [Mon, 16 Sep 2019 03:35:55 +0000 (04:35 +0100)] 
bpo-38117: Updated OpenSSL to 1.0.2t in macOS installer for 2.7.x. (GH-16171)

6 years ago[2.7] bpo-34155: Dont parse domains containing @ (GH-13079) (GH-16006)
Roberto C. Sánchez [Sat, 14 Sep 2019 17:26:38 +0000 (13:26 -0400)] 
[2.7] bpo-34155: Dont parse domains containing @ (GH-13079) (GH-16006)

This change skips parsing of email addresses where domains include a "@" character, which can be maliciously used since the local part is returned as a complete address.

(cherry picked from commit 8cb65d1381b027f0b09ee36bfed7f35bb4dec9a9)

Excludes changes to Lib/email/_header_value_parser.py, which did not
exist in 2.7.

Co-authored-by: jpic <jpic@users.noreply.github.com>
https://bugs.python.org/issue34155

6 years ago[2.7] bpo-35264: Modules/_ssl.c: fix build with OpenSSL 1.1.0 (GH-10570)
Alexandru Ardelean [Wed, 11 Sep 2019 17:23:28 +0000 (20:23 +0300)] 
[2.7] bpo-35264: Modules/_ssl.c: fix build with OpenSSL 1.1.0 (GH-10570)

Fixes a build error with OpenSSL 1.1.0. There is already code in the
`_ssl.c` that handles all the weird cases of the NPN config macros (with
various OpenSSL & LibreSSL versions).
That code will provide a HAVE_NPN variable, which should be used in the
rest of the code to check whether (or what) to compile regarding NPN.

This change adds HAVE_NPN in the remaining places where it should have been
placed.

Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
https://bugs.python.org/issue35264

6 years agocloses bpo-36712: Fix duplicate method in Lib/email/test/test_email_renamed.py. ...
Zackery Spytz [Wed, 11 Sep 2019 13:26:07 +0000 (07:26 -0600)] 
closes bpo-36712: Fix duplicate method in Lib/email/test/test_email_renamed.py. (GH-14800)

6 years agocloses bpo-36711: Remove duplicate method in Lib/email/feedparser.py. (GH-14801)
Zackery Spytz [Wed, 11 Sep 2019 13:22:08 +0000 (07:22 -0600)] 
closes bpo-36711: Remove duplicate method in Lib/email/feedparser.py. (GH-14801)

6 years agobpo-37445: Include FORMAT_MESSAGE_IGNORE_INSERTS in FormatMessageW() calls (GH-15822)
Zackery Spytz [Wed, 11 Sep 2019 09:39:34 +0000 (03:39 -0600)] 
bpo-37445: Include FORMAT_MESSAGE_IGNORE_INSERTS in FormatMessageW() calls (GH-15822)

If FormatMessageW() is passed the FORMAT_MESSAGE_FROM_SYSTEM flag
without FORMAT_MESSAGE_IGNORE_INSERTS, it will fail if there are
insert sequences in the message definition.
(cherry picked from commit a656365)

6 years ago[2.7] bpo-36502: Update link to UAX GH-44, the Unicode doc on the UCD. (GH-15808)
Benjamin Peterson [Mon, 9 Sep 2019 17:09:58 +0000 (10:09 -0700)] 
[2.7] bpo-36502: Update link to UAX GH-44, the Unicode doc on the UCD. (GH-15808)

The link we have points to the version from Unicode 6.0.0, dated 2010.
There have been numerous updates to it since then:
  https://www.unicode.org/reports/tr44/GH-Modifications

Change the link to one that points to the current version. Also, use HTTPS..
(cherry picked from commit 64c6ac74e254d31f93fcc74bf02b3daa7d3e3f25)

Co-authored-by: Greg Price <gnprice@gmail.com>
6 years ago[2.7] bpo-34410: Fix a crash in the tee iterator when re-enter it. (GH-15625) (GH...
Serhiy Storchaka [Mon, 9 Sep 2019 09:38:05 +0000 (12:38 +0300)] 
[2.7] bpo-34410: Fix a crash in the tee iterator when re-enter it. (GH-15625) (GH-15740)

RuntimeError is now raised in this case.
(cherry picked from commit 526a01467b3277f9fcf7f91e66c23321caa1245d)

6 years agocloses bpo-37965: Fix compiler warning of distutils CCompiler.test_function. (GH...
Miss Islington (bot) [Wed, 28 Aug 2019 17:36:18 +0000 (10:36 -0700)] 
closes bpo-37965: Fix compiler warning of distutils CCompiler.test_function. (GH-15560)

https://bugs.python.org/issue37965

https://bugs.python.org/issue37965

Automerge-Triggered-By: @benjaminp
(cherry picked from commit 55aabee07501e1468082b3237620e4ecd75c5da6)

Co-authored-by: Anonymous Maarten <madebr@users.noreply.github.com>
6 years agobpo-34521: Fix FD transfer in multiprocessing on FreeBSD (GH-15422)
Victor Stinner [Fri, 23 Aug 2019 13:00:38 +0000 (14:00 +0100)] 
bpo-34521: Fix FD transfer in multiprocessing on FreeBSD (GH-15422)

Fix file descriptors transfer in multiprocessing on FreeBSD: use
CMSG_SPACE() rather than CMSG_LEN(); see RFC 3542.

6 years agotest_multiprocessing removes temporary files (GH-15421)
Victor Stinner [Fri, 23 Aug 2019 12:59:40 +0000 (13:59 +0100)] 
test_multiprocessing removes temporary files (GH-15421)

(cherry picked from commit d0b10a64351069aa9246d40cb8bd207cc9209cee)

6 years agobpo-37823: Fix open() link in telnetlib doc (GH-15281)
Miss Islington (bot) [Wed, 21 Aug 2019 11:52:05 +0000 (04:52 -0700)] 
bpo-37823: Fix open() link in telnetlib doc (GH-15281)

Fixed wrong link to Telnet.open() method in telnetlib documentation.
(cherry picked from commit e0b6117e2723327d6741d0aa599408514add5b30)

Co-authored-by: Michael Anckaert <michael.anckaert@sinax.be>
6 years agobpo-35518: Skip test that relies on a deceased network service. (GH-15349)
Miss Islington (bot) [Wed, 21 Aug 2019 04:12:18 +0000 (21:12 -0700)] 
bpo-35518: Skip test that relies on a deceased network service. (GH-15349)

If this service had thoroughly vanished, we could just ignore the
test until someone gets around to either recreating such a service
or redesigning the test to somehow work locally.  The
`support.transient_internet` mechanism catches the failure to
resolve the domain name, and skips the test.

But in fact the domain snakebite.net does still exist, as do its
nameservers -- and they can be quite slow to reply.  As a result
this test can easily take 20-30s before it gets auto-skipped.

So, skip the test explicitly up front.
(cherry picked from commit 5b95a1507e349da5adae6d2ab57deac3bdd12f15)

Co-authored-by: Greg Price <gnprice@gmail.com>
6 years ago[2.7] bpo-37730: Fix usage of NotImplemented instead of NotImplementedError in docs... 15162/head
David H [Mon, 5 Aug 2019 09:19:26 +0000 (10:19 +0100)] 
[2.7] bpo-37730: Fix usage of NotImplemented instead of NotImplementedError in docs. (GH-15062). (GH-15133)

(cherry picked from commit ed5e8e06cbf766e89d6c58a882ee024abb5b2ed7)

Co-authored-by: David H <dheiberg@mozilla.com>
6 years agocloses bpo-37675: Use pkgutil.iter_modules to find fixers in a package rather than...
Miss Islington (bot) [Wed, 24 Jul 2019 23:59:31 +0000 (16:59 -0700)] 
closes bpo-37675: Use pkgutil.iter_modules to find fixers in a package rather than listdir. (14942)

(cherry picked from commit 93e8aa62cfd0a61efed4a61a2ffc2283ae986ef2)

Co-authored-by: Benjamin Peterson <benjamin@python.org>
6 years agobpo-34369: make kqueue.control() docs better reflect that timeout is positional-only...
Miss Islington (bot) [Thu, 11 Jul 2019 14:17:08 +0000 (07:17 -0700)] 
bpo-34369: make kqueue.control() docs better reflect that timeout is positional-only (GH-9499)

(cherry picked from commit 79042ac4348ccc09344014f20dd49401579f8795)

Co-authored-by: Tal Einat <taleinat@gmail.com>
6 years agoDocument default parameter of .seek() in the signature. (GH-14691)
Miss Islington (bot) [Thu, 11 Jul 2019 02:49:59 +0000 (19:49 -0700)] 
Document default parameter of .seek() in the signature. (GH-14691)

(cherry picked from commit 2a3d4d9c53dd4831c3ecf56bc7c4a289c33030d6)

Co-authored-by: Benjamin Peterson <benjamin@python.org>
6 years agobpo-37513: Change ValueError to TypeError in an example in ctypes doc (GH-14615) 14689/head
Miss Islington (bot) [Sun, 7 Jul 2019 15:59:14 +0000 (08:59 -0700)] 
bpo-37513: Change ValueError to TypeError in an example in ctypes doc (GH-14615)

(cherry picked from commit f6cdd3ff687ebbf8209d793a18a042ea495c4aeb)

Co-authored-by: Hai Shi <shihai1992@gmail.com>
6 years agobpo-37487: Fix PyList_GetItem index description. (GH-14623) (GH-14626)
Miss Islington (bot) [Sat, 6 Jul 2019 21:55:41 +0000 (14:55 -0700)] 
bpo-37487: Fix PyList_GetItem index description. (GH-14623) (GH-14626)

0 is a legal index.
(cherry picked from commit f8709e804d16ec5d44b1d2f00d59a0f78df7b792)

Co-authored-by: Terry Jan Reedy <tjreedy@udel.edu>
6 years agobpo-37149: Replace dead link for online Tkinter reference (GH-14616)
Miss Islington (bot) [Sat, 6 Jul 2019 07:04:53 +0000 (00:04 -0700)] 
bpo-37149: Replace dead link for online Tkinter reference  (GH-14616)

Also fix a name misspelling.
(cherry picked from commit 45bc61b97178b27ae05bd3eb95481bf0325795bb)

Co-authored-by: Terry Jan Reedy <tjreedy@udel.edu>
6 years agoPut pyexpatns.h include back. bpo-37437 (GH-14539) 14591/head
Miss Islington (bot) [Tue, 2 Jul 2019 05:38:31 +0000 (22:38 -0700)] 
Put pyexpatns.h include back. bpo-37437 (GH-14539)

(cherry picked from commit 2cd07920bb7d2d319999394092190f37935dc421)

Co-authored-by: Benjamin Peterson <benjamin@python.org>
6 years agobpo-19960: Fix building of zlib on macOS without installed headers (GH-14257)
Ned Deily [Mon, 1 Jul 2019 23:15:09 +0000 (19:15 -0400)] 
bpo-19960: Fix building of zlib on macOS without installed headers (GH-14257)

When building 2.7 on macOS without system header files installed in
``/usr/include``, a few extension modules dependent on system-supplied
third-party libraries were not being built, most notably zlib.
This situation arose in the past when building without the Command
Line Tools and the option to install header files in the traditional
system locations (like /usr/include).  As of macOS 10.14, the
header files are only available in an SDK so the problem addressed
here affects most 2.7 builds.

6 years ago[2.7] bpo-30754: Document textwrap.dedent blank line behavior. (GH-14469) (GH-14475)
Miss Islington (bot) [Sun, 30 Jun 2019 04:41:55 +0000 (21:41 -0700)] 
[2.7] bpo-30754: Document textwrap.dedent blank line behavior. (GH-14469) (GH-14475)

* Added documentation for textwrap.dedent behavior.
(cherry picked from commit eb97b9211e7c99841d6cae8c63893b3525d5a401)

Co-authored-by: tmblweed <tmblweed@users.noreply.github.com>
https://bugs.python.org/issue30754

6 years ago[2.7] bpo-37329: valgrind: ignore _PyWarnings_Init false alarms (GH-14202)
Victor Stinner [Fri, 28 Jun 2019 16:13:33 +0000 (18:13 +0200)] 
[2.7] bpo-37329: valgrind: ignore _PyWarnings_Init false alarms (GH-14202)

_PyWarnings_Init() only allocates memory once at startup but it is
not released at exit. Ignore this issue to be able to catch other
bugs more easily.

6 years agocloses bpo-37437: Update vendorized expat to 2.2.7. (GH-14436)
Miss Islington (bot) [Fri, 28 Jun 2019 04:24:52 +0000 (21:24 -0700)] 
closes bpo-37437: Update vendorized expat to 2.2.7. (GH-14436)

(cherry picked from commit 3b03b09fc94425915c5b1225e9200a3a95bc827b)

Co-authored-by: Benjamin Peterson <benjamin@python.org>
6 years ago[2.7] bpo-37411: Rewrite test_wsgiref.testEnviron() (GH-14394) (GH-14404)
Miss Islington (bot) [Wed, 26 Jun 2019 20:54:27 +0000 (13:54 -0700)] 
[2.7] bpo-37411: Rewrite test_wsgiref.testEnviron() (GH-14394) (GH-14404)

Fix test_wsgiref.testEnviron() to no longer depend on the environment
variables (don't fail if "X" variable is set).

testEnviron() now overrides os.environ to get a deterministic
environment. Test full TestHandler.environ content: not only a few
selected variables.
(cherry picked from commit 5150d327924959639215ed0a78feffc0d88258da)

Co-authored-by: Victor Stinner <vstinner@redhat.com>
6 years agobpo-37396, PCbuild: Include "_d" in "Killing any running ..." message (GH-14370)
Victor Stinner [Tue, 25 Jun 2019 11:37:16 +0000 (13:37 +0200)] 
bpo-37396, PCbuild: Include "_d" in "Killing any running ..." message (GH-14370)

Add $(PyDebugExt) in "Killing any running python$(PyDebugExt).exe
instances...".

6 years agobpo-35360: Update Windows builds to use SQLite 3.28.0 (GH-14182)
animalize [Mon, 24 Jun 2019 23:27:03 +0000 (07:27 +0800)] 
bpo-35360: Update Windows builds to use SQLite 3.28.0 (GH-14182)

6 years agobpo-37124: Fix reference leak in test_msilib (GH-13750) (GH-14340)
Victor Stinner [Mon, 24 Jun 2019 14:09:49 +0000 (16:09 +0200)] 
bpo-37124: Fix reference leak in test_msilib (GH-13750) (GH-14340)

(cherry picked from commit c0295dba259accc4b247beb22a0b2cc2f31d9850)

6 years agobpo-37359: Add --cleanup option to python3 -m test (GH-14332) (GH-14333)
Victor Stinner [Mon, 24 Jun 2019 11:21:18 +0000 (13:21 +0200)] 
bpo-37359: Add --cleanup option to python3 -m test (GH-14332) (GH-14333)

* regrtest: Add --cleanup option to remove "test_python_*" directories
  of previous failed test jobs.
* Add "make cleantest" to run "python -m test --cleanup".

(cherry picked from commit 47fbc4e45b35b3111e2d947a66490a43ac21d363)

6 years agoImprove threading.daemon docstring (GH-14278)
Miss Islington (bot) [Sun, 23 Jun 2019 19:08:28 +0000 (12:08 -0700)] 
Improve threading.daemon docstring (GH-14278)

Rephrase and clarify that "the entire Python program exits when only daemon threads are left". This matches the documentation at https://docs.python.org/3/library/threading.htmlGH-thread-objects.
(cherry picked from commit bb110cc2ed81447fb48805f31146cf31323a8fc3)

Co-authored-by: mbarkhau <mbarkhau@gmail.com>
6 years agobpo-37362: test_gdb now ignores stderr (GH-14287) (GH-14297)
Victor Stinner [Fri, 21 Jun 2019 21:58:53 +0000 (23:58 +0200)] 
bpo-37362: test_gdb now ignores stderr (GH-14287) (GH-14297)

test_gdb no longer fails if it gets an "unexpected" message on
stderr: it now ignores stderr. The purpose of test_gdb is to test
that python-gdb.py commands work as expected, not to test gdb.

(cherry picked from commit e56a123fd0acaa295a28b98d2e46d956b97d1263)

6 years agobpo-36231: Support building on macOS without /usr/include (GH-13773) (GH-14256)
Ned Deily [Thu, 20 Jun 2019 05:59:54 +0000 (01:59 -0400)] 
bpo-36231: Support building on macOS without /usr/include (GH-13773) (GH-14256)

6 years ago[2.7] Fix name of '\0'. (GH-14225)
Benjamin Peterson [Wed, 19 Jun 2019 06:03:35 +0000 (23:03 -0700)] 
[2.7] Fix name of '\0'. (GH-14225)

'\0' is the NUL byte not NULL..
(cherry picked from commit 7821b4c6d29933511d50bb42255e39790c6abf00)

6 years agobpo-34631: Updated OpenSSL to 1.0.2s in macOS installer. (GH-14198)
Ned Deily [Tue, 18 Jun 2019 10:48:53 +0000 (06:48 -0400)] 
bpo-34631: Updated OpenSSL to 1.0.2s in macOS installer. (GH-14198)

6 years ago[2.7] bpo-35360: Update macOS installer to use SQLite 3.28.0 (GH-14183)
animalize [Tue, 18 Jun 2019 09:59:53 +0000 (17:59 +0800)] 
[2.7] bpo-35360: Update macOS installer to use SQLite 3.28.0 (GH-14183)

6 years ago[2.7] Update link in colorsys docs to be https (GH-14062) (GH-14110)
Alex Gaynor [Tue, 18 Jun 2019 01:41:44 +0000 (21:41 -0400)] 
[2.7] Update link in colorsys docs to be https (GH-14062) (GH-14110)

6 years agobpo-34631: Updated OpenSSL to 1.0.2s in Windows installer (GH-14161)
Steve Dower [Mon, 17 Jun 2019 16:33:11 +0000 (09:33 -0700)] 
bpo-34631: Updated OpenSSL to 1.0.2s in Windows installer (GH-14161)

6 years ago[2.7] Fix 2.7 test -R test_IDLE failure on Windows (GH-13958)
Terry Jan Reedy [Sun, 16 Jun 2019 20:36:23 +0000 (16:36 -0400)] 
[2.7] Fix 2.7 test -R test_IDLE failure on Windows (GH-13958)

Cherry-picked from 66d47da.

6 years ago[2.7] bpo-35647: Fix path check in cookiejar. (GH-11436) (GH-13427)
Xtreak [Sat, 15 Jun 2019 16:29:29 +0000 (21:59 +0530)] 
[2.7] bpo-35647: Fix path check in cookiejar. (GH-11436) (GH-13427)

6 years ago[2.7] bpo-35121: prefix dot in domain for proper subdomain validation (GH-10258)...
Xtreak [Sat, 15 Jun 2019 15:29:43 +0000 (20:59 +0530)] 
[2.7] bpo-35121: prefix dot in domain for proper subdomain validation (GH-10258) (GH-13426)

This is a manual backport of ca7fe5063593958e5efdf90f068582837f07bd14 since 2.7 has `http.cookiejar` in `cookielib`

https://bugs.python.org/issue35121

6 years ago[2.7] bpo-36742: Fix urlparse.urlsplit() error message for Unicode URL (GH-13937)
Victor Stinner [Tue, 11 Jun 2019 10:45:35 +0000 (12:45 +0200)] 
[2.7] bpo-36742: Fix urlparse.urlsplit() error message for Unicode URL (GH-13937)

If urlparse.urlsplit() detects an invalid netloc according to NFKC
normalization, the error message type is now str rather than unicode,
and use repr() to format the URL, to prevent <exception str() failed>
when display the error message.

6 years ago[2.7] bpo-34836: fix test_default_ecdh_curve, needs no tlsv1.3. (GH-9626)
Dimitri John Ledkov [Sun, 9 Jun 2019 05:44:57 +0000 (06:44 +0100)] 
[2.7] bpo-34836: fix test_default_ecdh_curve, needs no tlsv1.3. (GH-9626)

Signed-off-by: Dimitri John Ledkov <xnox@ubuntu.com>
https://bugs.python.org/issue34836