David Benjamin [Fri, 22 Jul 2022 00:38:15 +0000 (17:38 -0700)]
Fix typo in PROTOCOL_TO_TLS_VERSION in test_ssl (GH-95119)
This appears to be a typo. It causes try_protocol_combo to try to turn
on SSL 3.0 when testing PROTOCOL_SSLv23 (aka PROTOCOL_TLS), which
doesn't make any sense. Fix it to be PROTOCOL_SSLv3.
Without this, try_protocol_combo is actually setting
context.minimum_version to SSLv3 when called as
try_protocol_combo(ssl.PROTOCOL_TLS, ssl.PROTOCOL_TLS, True)
One would think this causes a no-ssl3 OpenSSL build to fail, but OpenSSL
forgot to make SSL_CTX_set_min_proto_version(SSL3_VERSION) does not
notice no-ssl3, so this typo has gone undetected. But we should still
fix the typo because, presumably, a future version of OpenSSL will
remove SSL 3.0 and do so more thoroughly, at which point this will
break.
David Benjamin [Thu, 21 Jul 2022 18:44:30 +0000 (11:44 -0700)]
gh-95095: Use SSL_CTX_get_max_proto_version instead of SSL_CTX_ctrl (GH-95096)
The wrapper macros are more readable and match the form recommended in
the OpenSSL documentation. They also slightly less error-prone, as the
mapping of arguments to SSL_CTX_ctrl is not always clear. (Though in
this case it's straightforward.)
https://www.openssl.org/docs/man1.1.1/man3/SSL_CTX_get_max_proto_version.html
Ned Batchelder [Wed, 20 Jul 2022 21:47:43 +0000 (14:47 -0700)]
docs: use 'recursively' in the description of rglob, and mention globs in the os equivalences (GH-94954)
The r in `rglob` stands for "recursively", so use the word in the description. Also, glob and rglob can usefully be mentioned as the pathlib equivalent of os.walk.
gh-95007: Remove the NoneType return converter (GH-95019)
It has confusing semantic which does not provide any benefit (the
only difference is that you should write "return Py_None" instead
of "Py_RETURN_NONE"), it is not currently used, and it is broken.
Thomas Grainger [Sun, 17 Jul 2022 01:07:53 +0000 (02:07 +0100)]
gh-91181: drop support for bytes on sys.path (GH-31934)
Support for bytes broke sometime between Python 3.2 and 3.6 and has been broken ever since. Trying to bring back supports is surprisingly difficult in the face of -b and checking for keys in sys.path_importer_cache. Since the support was broken for so long, trying to overcome the difficulty of bringing back the support has been deemed not worth it.
Co-authored-by: Eryk Sun <eryksun@gmail.com> Co-authored-by: Brett Cannon <brett@python.org>
Chris Adams [Fri, 15 Jul 2022 22:09:27 +0000 (18:09 -0400)]
Add additional pointers to pathlib's mapping to os.path functions (#94828)
* Add additional pointers to pathlib's mapping to os.path functions
os.path.splitext has a somewhat quirky signature since it mixes the path and filename components but I wanted the documentation to mention `PurePath.stem` as the natural counterpart to `PurePath.suffix` for the common use of `os.path.splitext` to turn "file.py" into "file" and "py".
Technically this could have some discussion of how to handle the parent directory hierarchy but that seems a bit out of keeping with the spirit of this table so I omitted mentioning `PurePath.parents` here.
gh-94821: Fix autobind of empty unix domain address (GH-94826)
When binding a unix socket to an empty address on Linux, the socket is
automatically bound to an available address in the abstract namespace.
>>> s = socket.socket(socket.AF_UNIX, socket.SOCK_STREAM)
>>> s.bind("")
>>> s.getsockname()
b'\x0075499'
Since python 3.9, the socket is bound to the one address:
>>> s.getsockname()
b'\x00'
And trying to bind multiple sockets will fail with:
Traceback (most recent call last):
File "/home/nsoffer/src/cpython/Lib/test/test_socket.py", line 5553, in testAutobind
s2.bind("")
OSError: [Errno 98] Address already in use
Added 2 tests:
- Auto binding empty address on Linux
- Failing to bind an empty address on other platforms
gh-94816: Improve coverage of decode_linetable (GH-94853)
This makes calls to co_lnotab to exercise this code, as well
as generating synthetically large code to exercise the corner
cases where line numbers need multiple bytes.
Petr Viktorin [Tue, 12 Jul 2022 15:06:05 +0000 (17:06 +0200)]
gh-94751: Install, import and run the test C++ extension (MVP) (GH-94754)
This is a quick-and-dirty way to run the C++ tests.
It can definitely be improved in the future, but it should fail when things go wrong.
- Run test functions on import (yes, this can definitely be improved)
- Fudge setuptools metadata (name & version) to make the extension installable
- Install and import the extension in test_cppext
John Belmonte [Mon, 11 Jul 2022 06:40:53 +0000 (15:40 +0900)]
gh-93883: elide traceback indicators when possible (#93994)
* gh-93883: elide traceback indicators when possible
Elide traceback column indicators when the entire line of the
frame is implicated. This reduces traceback length and draws
even more attention to the remaining (very relevant) indicators.
Example:
```
Traceback (most recent call last):
File "query.py", line 99, in <module>
bar()
File "query.py", line 66, in bar
foo()
File "query.py", line 37, in foo
magic_arithmetic('foo')
File "query.py", line 18, in magic_arithmetic
return add_counts(x) / 25
^^^^^^^^^^^^^
File "query.py", line 24, in add_counts
return 25 + query_user(user1) + query_user(user2)
^^^^^^^^^^^^^^^^^
File "query.py", line 32, in query_user
return 1 + query_count(db, response['a']['b']['c']['user'], retry=True)
~~~~~~~~~~~~~~~~~~^^^^^
TypeError: 'NoneType' object is not subscriptable
```
Rather than going out of our way to provide indicator coverage
in every traceback test suite, the indicator test suite should
be responible for sufficient coverage (e.g. by adding a basic
exception group test to ensure that margin strings are covered).