Victor Stinner [Sat, 21 Jul 2018 00:06:16 +0000 (02:06 +0200)]
bpo-34170: Rework _PyCoreConfig_Read() to avoid side effect (GH-8353)
Rework _PyCoreConfig_Read() function which *reads* core configuration
to not *modify* the path configuration.
A new _PyCoreConfig_SetPathConfig() function now recreates the path
configuration from the core configuration. This function is now
called very late in _Py_InitializeCore(), just before calling
initimport().
Changes:
* Add _PyCoreConfig.dll_path
* Py_SetPath() now fails with a fatal python error on memory
allocation failure.
* Rename _PyPathConfig_Calculate() to _PyPathConfig_Calculate_impl()
* Replace _PyPathConfig_Init() with _PyPathConfig_Calculate(): the
function now requires a _PyPathConfig
* Add _PyPathConfig_SetGlobal() to set the _Py_path_config global
variable.
* Add _PyCoreConfig_InitPathConfig(): compute the path configuration
* Add _PyCoreConfig_SetPathConfig(): set path configuration from core
configuration
* Rename wstrlist_append() to _Py_wstrlist_append()
* _Py_wstrlist_append() now handles integer overflow.
Victor Stinner [Fri, 20 Jul 2018 15:34:23 +0000 (17:34 +0200)]
bpo-34008: Allow to call Py_Main() after Py_Initialize() (GH-8043)
Py_Main() can again be called after Py_Initialize(), as in Python
3.6. The new configuration is ignored, except of
_PyMainInterpreterConfig.argv which is used to update sys.argv.
Increase the timeout: give timeout x 4 instead of timeout x 2 to
threads to wait until the Event is set, but reduce the sleep from 500
ms to 250 ms. So the test should be more reliable and faster!
On Windows, sometimes test_signal.test_warn_on_full_buffer() fails to
fill the socketpair buffer. In that case, the C signal handler
succeed to write into the socket, it doesn't log the expected send
error, and so the test fail.
On Windows, the test now uses a timeout of 50 ms to fill the
socketpair buffer to fix this race condition.
Other changes:
* Begin with large chunk size to fill the buffer to speed up the
test.
* Add error messages to assertion errors to more easily identify
which assertion failed.
* Don't set the read end of the socketpair as non-blocking.
test_signal.test_socket(): On Windows, sometimes even if the C signal handler
succeed to write the signal number into the write end of the socketpair, the
test fails with a BlockingIOError on the non-blocking read.recv(1) because the
read end of the socketpair didn't receive the byte yet.
Fix the race condition on Windows by setting the read end as blocking.
The test failed on AMD64 Debian root 3.x buildbot because the busy
loop of 100 ms only increased time.thread_time() by 19.9 ms which is
smaller than 20 ms. Modify the test to tolerate a delta of at least
15 ms instead of 20 ms.
Nick Coghlan [Tue, 10 Jul 2018 13:04:29 +0000 (23:04 +1000)]
bpo-34009: Expand on platform support changes (GH-8022)
Many users won't realise the implications the OpenSSL 1.0.2
minimum version requirement has when it comes to Ubuntu
14.04 and Debian 8, so expand on that in the platform support
section.
Also explicitly note the non-ASCII-based locale requirement for
full Unicode text handling support on non-Windows systems.
bpo-34066: Disabled interruption before SETUP_WITH and BEFORE_ASYNC_WITH. (GH-8159)
This will prevent emitting a resource warning when the execution was
interrupted by Ctrl-C between calling open() and entering a 'with' block
in "with open()".
Andrew Kuchling [Sat, 7 Jul 2018 21:36:23 +0000 (17:36 -0400)]
bpo-28626: rearrange discussion of output formatting to encourage f-strings (#6036)
The 'output formatting' section of the tutorial talks a lot about manual formatting with things like .rjust() and .zfill(), with only a passing reference to 3.6's new f-strings.
This doesn't drop all of the old material, but it does rearrange the topics into a more modern order: f-strings first, discussing formatting specifiers a bit; then calling .format(); finally manual formatting with .ljust().
closes bpo-34056: Always return bytes from _HackedGetData.get_data(). (GH-8130)
* Always return bytes from _HackedGetData.get_data().
Ensure the imp.load_source shim always returns bytes by reopening the file in
binary mode if needed. Hash-based pycs have to receive the source code in bytes.
It's tempting to change imp.get_suffixes() to always return 'rb' as a mode, but
that breaks some stdlib tests and likely 3rdparty code, too.
bpo-33899: Make tokenize module mirror end-of-file is end-of-line behavior (GH-7891)
Most of the change involves fixing up the test suite, which previously made
the assumption that there wouldn't be a new line if the input didn't end in
one.
tarfile._Stream has two buffer for compressed and uncompressed data.
Those buffers are not aligned so unnecessary bytes slicing happens
for every reading chunks.
This commit bypass compressed buffering.
In this benchmark [1], user time become 250ms from 300ms.
subprocess.Popen now copies the startupinfo argument to leave it
unchanged: it will modify the copy, so that the same STARTUPINFO
object can be used multiple times.
Result of function interpreter_requires_environment() depends on os.environ.
This was not covered by the tests, leading to fail when PYTHONHOME was set.
Fix test_forkserver_sigkill() of test_multiprocessing_forkserver:
give more time to the first child process to complete, double the
sleep in the parent process.
Reduce also the child process sleep from 1000 ms to 500 ms, to not change
the total duration of the test.
During buffered read, use a list followed by join instead of extending a bytes object.
This is how it was done before but changed in commit b506dc32c1a.