As of Xcode 7, SDKs for Apple platforms now include textual-format stub
libraries whose file names have a .tbd extension rather than the
standard OS X .dylib extension. The Apple compiler tool chain handles
these stub libraries transparently and the installed system shared libraries
are still .dylibs. However, the new stub libraries cause problems for
third-party programs that support building with Apple SDKs and make
build-time decisions based on the presence or paths of system-supplied
shared libraries in the SDK. In particular, building Python itself with
an SDK fails to find system-supplied libraries during setup.py's build of
standard library extension modules. The solution is to have
find_library_file() in Distutils search for .tbd files, along with
the existing types (.a, .so, and .dylib). Patch by Tim Smith.
* There are only two base-64 alphabets defined by the RFCs, not three
* Due to the internal translation, plus (+) and slash (/) are never discarded
* standard_ and urlsafe_b64decode() discard characters as well
Also update the doc strings to clarify data types, based on revision
92760d2edc9e, correct the exception raised by b16decode(), and correct the
parameter name for the base-85 functions.
importlib.util.LazyLoader.
The class was checking its argument as to whether its implementation
of create_module() came directly from importlib.abc.Loader. The
problem is that the classes coming from imoprtlib.machinery do not
directly inherit from the ABC as they come from _frozen_importlib.
Because the documentation has always said that create_module() was
ignored, the check has simply been removed.
* The Windows-specific binary notice was probably a Python 2 thing
* Make it more obvious gettarinfo() is based on stat(), and that non-ordinary
files may need special care
* The file name must be text; suggest dummy arcname as a workaround
* Indicate TarInfo may be used directly, not just via gettarinfo()
Microbenchmarks show 2-2.5x improvement. Built-in 'divmod' function
is now also ~10% faster.
-m timeit -s "x=22331" "x//2;x//-3;x//4;x//5;x//-6;x//7;x//8;x//-99;x//100;"
with patch: 0.321 without patch: 0.633
-m timeit -s "x=22331" "x%2;x%3;x%-4;x%5;x%6;x%-7;x%8;x%99;x%-100;"
with patch: 0.224 without patch: 0.66
Big thanks to Serhiy Storchaka, Mark Dickinson and Victor Stinner for
thorow code reviews and algorithms improvements.
The compile ignores constant statements and emit a SyntaxWarning warning.
Don't emit the warning for string statement because triple quoted string is a
common syntax for multiline comments.
Don't emit the warning on ellipis neither: 'def f(): ...' is a legit syntax for
abstract functions.
Changes:
* test_ast: ignore SyntaxWarning when compiling test statements. Modify
test_load_const() to use assignment expressions rather than constant
expression.
* test_code: add more kinds of constant statements, ignore SyntaxWarning when
testing that the compiler removes constant statements.
* test_grammar: ignore SyntaxWarning on the statement "1"
* Constant statements will be ignored and the compiler will emit a
SyntaxWarning.
* Replace constant statement (ex: "1") with an expression statement
(ex: "x=1").
* test_traceback: use context manager on the file.
Issue #26204.
Fixed a crash when unpickle the functools.partial object with wrong state.
Fixed a leak in failed functools.partial constructor.
"args" and "keywords" attributes of functools.partial have now always types
tuple and dict correspondingly.
Fixed a crash when unpickle the functools.partial object with wrong state.
Fixed a leak in failed functools.partial constructor.
"args" and "keywords" attributes of functools.partial have now always types
tuple and dict correspondingly.
* test_contains() did not override anything
* test_expandtabs/upper/lower() in FixedStringTest were masking usable tests
in string_tests. These tests now get run for bytearray() and bytes().
* test_expandtabs/upper/lower() in buffer_tests were only run on bytearray()
and are redundant with string_tests
Test test_wrong_cert() runs a server that rejects the client's certificate,
so ECONNRESET is reasonable in addition to SSLError. On the other hand, the
other three tests don't even need to run a server because they are just
testing the parsing of invalid certificate files.
Also fix a ResourceWarning by closing the wrapped socket.
This will hopefully make maintenance of venv documentation easier.
For example, see commits a4f0d76af176 and 5764cc02244d.
This patch has been reviewed by Vinaj Sajip, the maintainer of
venv module.
Testing for a non-existing certificate file is already done in test_errors().
Copy wrongcert.pem from Python 2 and use it to test the behaviour with a
mismatched certificate.
* Issue #25234: Skip test_eintr.test_open() under OS X to avoid hanging
* Issue #25868: Try to make test_eintr.test_sigwaitinfo() more reliable
especially on slow buildbots. Use a pipe to synchronize the parent and the
child processes.
Issue #26146: Add a new kind of AST node: ast.Constant. It can be used by
external AST optimizers, but the compiler does not emit directly such node.
An optimizer can replace the following AST nodes with ast.Constant:
* ast.NameConstant: None, False, True
* ast.Num: int, float, complex
* ast.Str: str
* ast.Bytes: bytes
* ast.Tuple if items are constants too: tuple
* frozenset
Update code to accept ast.Constant instead of ast.Num and/or ast.Str:
* compiler
* docstrings
* ast.literal_eval()
* Tools/parser/unparse.py
with no known parent package.
Previously SystemError was raised if the parent package didn't exist
(e.g., __package__ was set to '').
Thanks to Florent Xicluna and Yongzhi Pan for reporting the issue.
In a previous change, __spec__.parent was prioritized over
__package__. That is a backwards-compatibility break, but we do
eventually want __spec__ to be the ground truth for module details. So
this change reverts the change in semantics and instead raises an
ImportWarning when __package__ != __spec__.parent to give people time
to adjust to using spec objects.
Issue #25843: When compiling code, don't merge constants if they are equal but
have a different types. For example, "f1, f2 = lambda: 1, lambda: 1.0" is now
correctly compiled to two different functions: f1() returns 1 (int) and f2()
returns 1.0 (int), even if 1 and 1.0 are equal.
Add a new _PyCode_ConstantKey() private function.
Issue #25843: When compiling code, don't merge constants if they are equal but
have a different types. For example, "f1, f2 = lambda: 1, lambda: 1.0" is now
correctly compiled to two different functions: f1() returns 1 (int) and f2()
returns 1.0 (int), even if 1 and 1.0 are equal.
Add a new _PyCode_ConstantKey() private function.
Issue #26099: The site module now writes an error into stderr if sitecustomize
module can be imported but executing the module raise an ImportError. Same
change for usercustomize.
Issue #26107: The format of the co_lnotab attribute of code objects changes to
support negative line number delta.
Changes:
* assemble_lnotab(): if line number delta is less than -128 or greater than
127, emit multiple (offset_delta, lineno_delta) in co_lnotab
* update functions decoding co_lnotab to use signed 8-bit integers
- dis.findlinestarts()
- PyCode_Addr2Line()
- _PyCode_CheckLineNumber()
- frame_setlineno()
* update lnotab_notes.txt
* increase importlib MAGIC_NUMBER to 3361
* document the change in What's New in Python 3.6
* cleanup also PyCode_Optimize() to use better variable names
The previous test relied on a remote server, which currently seems to be
shutting the connection down once TLS has been set up, causing an EOFError.
Now the test is implemented using a minimal NNTP server running in a
background thread.
Issue #26100:
* Add subprocess._optim_args_from_interpreter_flags()
* Add test.support.optim_args_from_interpreter_flags()
* Use new functions in distutils, test_cmd_line_script, test_compileall and
test_inspect
The change enables test_details() test of test_inspect when -O or -OO command
line option is used.
Issue #26101: Exclude Lib/test/ from sys.path in test_compilepath(). The
directory contains invalid Python files like Lib/test/badsyntax_pep3120.py,
whereas the test ensures that all files can be compiled.
Otherwise, GDB seems to affect the terminal's foreground process group,
interfering with test_ioctl, which does not expect the foreground process to
change during the test. This change also solves the problem of the tests
being stopped in the shell if test_gdb is run twice in parallel.
not defined for a relative import.
This is the start of work to try and clean up import semantics to rely
more on a module's spec than on the myriad attributes that get set on
a module. Thanks to Rose Ames for the patch.
Previously zipimport mistakenly limited namespace support to only the
top-level of the zipfile when it should have supported an arbitrary
depth.
Thanks to Phil Connel for the bug report and initial patch and Mike
Romberg for the final patch.
This is instead of svn.python.org, whose certificate recently expired, and
whose new certificate uses a different root certificate.
The certificate used at the pythontest server was modifed to set the "basic
constraints" CA flag. This flag seems to be required for test_get_ca_certs_
capath() to work (in Python 3.4+).
Added the new self-signed certificate to capath with the following commands:
cp Lib/test/{selfsigned_pythontestdotnet.pem,capath/}
c_rehash -v Lib/test/capath/
c_rehash -v -old Lib/test/capath/
# Note the generated file names
cp Lib/test/capath/{selfsigned_pythontestdotnet.pem,0e4015b9.0}
mv Lib/test/capath/{selfsigned_pythontestdotnet.pem,ce7b8643.0}
The new server responds with "No route to host" when connecting to port 444.
Ned Deily
Jason R. Coombs
Serhiy Storchaka
Martin Panter
Brett Cannon
Berker Peksag
Yury Selivanov
Vinay Sajip
Charles-François Natali
Victor Stinner
Anish Shah
Eric V. Smith
Raymond Hettinger
Terry Jan Reedy
Senthil Kumaran
Donald Stufft
Benjamin Peterson
Steve Dower
Gregory P. Smith
Ethan Furman
Martin Panter
Barry Warsaw
doko@ubuntu.com